When will AGI arrive? Tracking 14 AI industry leaders' predictions from 2014 to 2026

Introduction

AI eradicates the vast majority of cancers
AI makes in 5–10 years the Nobel Prize–level discoveries that would have taken humans 50–100 years
AI puts 20% of white-collar workers out of a job
AI wipes out humanity
AI secures adequate food, clothing, and shelter for every human being, with no one having to work

Claims like these sound unrealistic, yet they have actually been made by the leaders of major AI companies such as Open AI, Google, and Anthropic. Some already have real-world precedents: Demis Hassabis of Google DeepMind won the Nobel Prize in Chemistry after using AI to solve the prediction of proteins' three-dimensional structures—a problem that had gone unsolved for more than 50 years—in roughly a decade of research, and large corporations such as Amazon, IBM, and Meta have carried out layoffs of thousands to tens of thousands of employees, many of them white-collar workers.

AI is now advancing at a staggering pace, and with the future so hard to predict, one useful anchor is what the CEOs and other leaders of AI companies have said. Elon Musk, for instance, declared in 2014 that "AI is more dangerous than nuclear weapons," and in 2019 that "AI will make jobs pointless. Probably the last job will be writing AI software, and then AI will write its own software." In 2026, these remarks feel remarkably close to reality.

This article looks at the major AI-related companies and organizes their leaders' statements chronologically. Tracing each person's remarks over time makes it easier to grasp their outlook on and stance toward AI, and lets us sketch several scenarios for what might happen over the next few years to a decade.

A major turning point came in 2014 with the publication of Nick Bostrom's Superintelligence: Paths, Dangers, Strategies, which raised the dangers posed by the emergence of AGI (Artificial General Intelligence: AI with human-level intelligence, capable of learning, applying knowledge, and solving problems across diverse fields). Accordingly, this article organizes the statements of each key figure from 2014 onward. Because the full material is extremely long, a summary overview precedes the year-by-year statements. The actual statements appear further down for those who are interested.

Who is covered

We cover the following major AI developers and their leaders. The titles given alongside each statement are those held at the time (including former positions).

Open AI

• Sam Altman
• Greg Brockman

Anthropic

• Dario Amodei
• Daniela Amodei

Google

• Sundar Pichai
• Demis Hassabis

xAI

• Elon Musk

Mistral AI

• Arthur Mensch

DeepSeek

• Liang Wenfeng

Meta

• Mark Zuckerberg
• Alexandr Wang

NVIDIA

• Jensen Huang

Microsoft

• Satya Nadella
• Mustafa Suleyman

 

Summary

Because the statements of each figure since 2014 run so long, they are presented in the sections that follow, for reference.
Here we present the summary first, organizing each figure's key statements, each company's stance, and the outlook for AI going forward.

Each company's stance

We organize each company's stance along the following three dimensions.

• AGI outlook: When the company expects AGI—artificial general intelligence, the next generation of AI with intelligence equal to or surpassing humans and able to handle diverse tasks flexibly—to arrive
• Open vs. closed strategy: The company's policy on how much of the models it develops (weights, code, training data, papers) to release publicly
• Political and geopolitical stance: The company's posture toward national governments, regulation, national security, export controls, and international cooperation

AGI outlook

• Accelerated outlook
  • Open AI: Arrives in 2028–30
  • Anthropic: Arrives in 2026–27
  • Google: Arrives in 2030–35
  • xAI: Arrives by 2030
• Cautious outlook
  • Microsoft: Rather than an AI that can do anything, build Humanist Superintelligence (AI that serves humans and operates under human control)
• Skeptical outlook
  • Meta: AGI is not a useful concept
  • Mistral AI: AGI is not a useful concept
• Silent
  • DeepSeek: Research first
  • NVIDIA: Neutral, as an infrastructure provider

Open vs. closed strategy

• Open
  • Mistral AI: Fully open
  • DeepSeek: Fully open (MIT)
• Partially open
  • Meta: Was fully open, but has changed course
  • Google: Closed, with AlphaFold released publicly
  • xAI: Partially open (Grok-1)
• Closed
  • Open AI: Closed API, with some open releases
  • Anthropic: Closed

Political and geopolitical stance

• U.S. sovereignty; cooperation with the U.S. government
  • Open AI: Aligned with the Trump administration
  • xAI: The most politicized AI company
  • Meta: AGI is not a useful concept
• In conflict with the U.S. government
  • Anthropic: Head-on confrontation with the Trump administration
• European sovereignty
  • Mistral AI: Close collaboration with the French government; reducing dependence on the U.S.
• Chinese sovereignty
  • DeepSeek: A symbol of China's AI sovereignty
• Keeping a distance from governments
  • Google: Global science first
  • Microsoft: Cooperation with the U.S., the EU, and national governments
• Neutral
  • NVIDIA: Opposed to global export controls

Key statements from each figure

Open AI

• Sam Altman (CEO)
  • "Technology brings both an equalization of capability and a concentration of wealth. Total assets grow, but end up concentrated in the hands of a few." (2017)
  • "We're not far from being able to eliminate poverty and provide universal healthcare. Enough food and a place to live will become basic human rights." (2017)
  • "The defining characteristic of the Intelligence Age will be massive prosperity." "Astounding triumphs—fixing the climate, establishing a space colony, and the discovery of all of physics—will eventually become commonplace." (2024)
  • "Access to AI should be a fundamental human right." (2025)
  • "If this technology becomes concentrated in a single company or country, it could lead to catastrophe." (2026)
  • "We still don't know how to think about superintelligence combined with a dictator or a totalitarian state." (2026)
• Greg Brockman (CTO)
  • "I'm taking time off through the end of the year—my first chance to relax in the nine years since co-founding OpenAI." "The mission is still far from complete. We have to build safe artificial general intelligence (AGI)." (2024)
  • "GPT-5.5 brings us one step closer to realizing OpenAI's 'super app.'" (2026)

Anthropic

• Dario Amodei (CEO)
  • "Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems." (2021)
  • "I think most people underestimate just how dramatic the benefits of AI could be, and I think they also underestimate how severe the risks could be." (2024)
  • "AI-enabled biology and medicine will allow us to compress the progress that human biologists would have achieved over the next 50–100 years into 5–10 years." (2024)
  • "AI could wipe out half of all entry-level white-collar jobs and spike unemployment to 10–20% in the next one to five years." (2025)
  • "My sense is that almost every decision I make is a real tightrope walk. If we don't build fast enough, the authoritarian countries could win." (2025)
  • "AI's goals and behavior misaligned with human intent (behavior including deception, blackmail, and scheming has already been observed in Anthropic's internal tests)" (2026)
  • "Destructive misuse of AI by individuals or small groups (especially concerns about biological weapons: 'people without specialized training will be able to build weapons of mass destruction')" (2026)
  • "AI agents will become the operating layer of Wall Street." (2026)
• Daniela Amodei (President)
  • "We're constantly thinking about how to make Anthropic a place where safety-minded people can work productively." (2024)
  • "The constraint on hiring isn't motivation; it's ability and cultural fit." (2024)

Google

• Sundar Pichai (Alphabet CEO)
  • "AI is probably the most important thing humanity has ever worked on. I think of it as something more profound than electricity or fire." (2018)
  • "We (Google) are standing at the dawn of a new era of computing." (2024)
  • "AI is too important not to regulate, and too important not to regulate well." (2024)
• Demis Hassabis (Google DeepMindCTO)
  • "AI has enormous potential to be amazing for humanity. It will really accelerate progress on solving disease and in all sorts of areas where we're currently making relatively slow progress." (2014)
  • "In 2026, we stand at a new inflection point, with AGI (artificial general intelligence) on the horizon—likely within the next five years." (2026)
  • "These general-purpose, foundation-model systems are improving in capability almost week by week." (2026)
  • "We also need to approach this with the understanding and humility that we don't yet have all the answers about how this technology will develop." (2026)

xAI

• Elon Musk (CEO)
  • "AI is more dangerous than nuclear weapons" (2014)
  • "With artificial intelligence, we are summoning the demon." (2014)
  • "AI doesn't have to be evil to destroy humanity. If AI has a goal and humanity just happens to be in the way, it will destroy humanity as a matter of course, without even thinking about it. No hard feelings." "The least scary future I can think of is one where we have at least democratized AI." (2018)
  • "AI will make jobs pointless. Probably the last job will be writing AI software, and then AI will write its own software." (2019)
  • "OpenAI has been transformed into a de facto closed-source subsidiary of Microsoft, the world's largest technology company." "OpenAI has abandoned its mission of developing artificial general intelligence (AGI) for the benefit of humanity." (2024)
  • "I've long believed that AI would become far smarter than humans and pose an existential threat to humanity, and that is now becoming reality. There is a 20% chance of AI 'annihilating' humanity." (2025)
  • "AI will surpass the intelligence of the smartest human by next year." (2026)
  • "AI will be smarter than all of humanity combined by 2029." (2026)

Mistral AI

• Arthur Mensch (CEO)
  • "Europe needs to reduce its dependence on the United States for foundational AI technology." (2024)
  • Criticized the AGI debate as tantamount to "creating God" (2025)
  • "Regulation should target applications, not foundation models." "Open-source models guarantee safety." (2026)
  • "Sounds the alarm about the risk of 'deskilling' (the loss of human skills)." (2026)

DeepSeek

• Liang Wenfeng (CEO)
  • "As its economy develops, China must gradually shift from being a beneficiary to being a contributor." (2023)
  • "We're done following in others' footsteps. Now it's our turn to lead." (2025)
  • "Innovation isn't purely business-driven; it requires curiosity and creative ambition." (2026)
  • "In disruptive technologies, closed-source moats are fleeting." (2026)
  • "Open source is cultural, not commercial." (2026)

Meta

• Mark Zuckerberg (CEO)
  • "Meta has a long history of open-sourcing its infrastructure and AI work, from PyTorch to React. Open source drives innovation." (2023)
  • "I expect that within the next 12 to 18 months, most of the code used for these efforts will be written by AI." (2025)
• Alexandr Wang (CTO)
  • In October 2025, when announcing 600 layoffs within MSL: "These cuts are intended to correct the bureaucratic bloat of the previous organization and move us toward a leaner structure that concentrates top talent." (2025)
  • "The AI race between the US and China is the most important competition of our generation." (2026)

NVIDIA

• Jensen Huang (CEO)
  • "In the future, we'll program in natural language. The computer will handle the rest." (2024)
  • "The ChatGPT moment for physical AI has arrived. The time has come for machines to start understanding the real world, reasoning about it, and acting in it." (2025)
  •  Responding to Amodei's assertion that "US tech sanctions will hold China back": "AI is a global technology and should not be weaponized in geopolitical struggles." (2026)  
  • "AI factories are the new infrastructure of the 21st century." (2026)
  • "Every company will become an AI company." (2026)

Microsoft

• Satya Nadella (Microsoft CEO)
  • "In the end, it won't be humans versus machines; it will be humans coexisting with machines." (2015)
  • "Bots are the new apps. People talking to people, people talking to digital assistants, people talking to bots, and digital assistants talking to bots. That's the world you're going to see over the next few years." (2016)
  • "Every company is becoming a software company, and every software company is becoming an AI company." "AI is the runtime that powers every application." (2022)
  • "AI must be designed for people." (2022)
  • "The big winners here won't be the technology companies. The winners will be the broader industries that make use of this resource—which, incidentally, is abundant." (2025)
• Mustafa Suleyman (Microsoft AI CEO)
  • "At Microsoft AI, we believe humans matter more than AI." (2025) 
  • "A superintelligence project should be about designing AI that remains subordinate to humans and keeps humanity at the top of the food chain." (2025)
  • "Within the next five years, every individual will have a highly personalized AI companion." (2026)
  • "AI is already superhuman." "You can't steer what you can't control." "Containment and alignment are essential preconditions—red lines. I think everyone in the industry has to get behind that idea." (2026)

Where AI Goes From Here

The picture can be summarized roughly as follows.

• 2026–2027: Autonomous agents enter practical use, executing complex multi-step tasks without human supervision (Altman: "agents will join the workforce in 2025"; Brockman: "GPT-5.5 plans and executes complex tasks"; Hassabis: "Gemini ushers in the agentic era")
• 2027–2028: Robots operating in the physical world arrive alongside the automation of half of all white-collar jobs, bringing social upheaval with unemployment at 10–20% (Altman: "robots performing tasks in the real world by 2027"; Amodei: "half of entry-level white-collar jobs will disappear")
• 2028–2030: AGI (human-level general intelligence) is achieved at multiple research organizations, and a "compressed 21st century" begins in medicine and scientific research (Hassabis: "a 50% chance of AGI within 5–10 years"; Amodei: "a country of geniuses in a datacenter"; Musk: "surpassing all of humanity by 2029")
• Early 2030s: The transition to superintelligence; intelligence and energy become "wildly abundant," and progress on climate change, cancer, and space expansion becomes routine (Altman: "abundant intelligence and energy in the 2030s"; Suleyman: "medical superintelligence will arrive first")
• Late 2030s and beyond: A fork in the road where the relationship between humanity and AI is fundamentally redefined—prosperity through "human-centered superintelligence," or a civilizational crisis caused by loss of control. Here the leaders' forecasts diverge sharply (Suleyman: "Humanist Superintelligence"; Musk: "a 20% risk of extinction"; Amodei: "on a knife's edge"; Altman: "the Gentle Singularity")

Closing Thoughts

We are living through a moment in which the progress of AI is forcing fundamental questions about society and what it means to be human. I intend to keep watching closely what AI industry leaders say and to let that inform my own next steps. Below, I have organized the statements year by year in chronological order. If you are interested, reading through them gives a clear picture of how we got here.

2014: Events, Positions, and Statements

Key Events

• January 26: Google acquires DeepMind for approximately $500M (Demis Hassabis, Shane Legg, Mustafa Suleyman)
• June: Facebook AI Research (FAIR) publishes the "DeepFace" paper — facial recognition reaches human-level accuracy above 97%
• June: Ian Goodfellow publishes the GAN (Generative Adversarial Networks) paper
• July: Nick Bostrom publishes Superintelligence: Paths, Dangers, Strategies, the starting point of the debate over existential risk from AI. Elon Musk, Bill Gates, Sam Altman, Stephen Hawking, and others publicly acknowledge its influence.
• December: Stephen Hawking tells the BBC, "The development of full artificial intelligence could spell the end of the human race."

Positions and Statements by Individual

OpenAI

• Sam Altman（Y Combinator CEO）
  • None of note
• Greg Brockman (CTO, Stripe)
  • None of note

People

• Dario Amodei (postdoctoral researcher at Princeton → Baidu)
  • None of note
• Daniela Amodei (risk management at Stripe)
  • None of note

Google

• Sundar Pichai (Chief of Product, Google)
  • None of note
• Demis Hassabis (CEO, DeepMind)
  • DeepMind's mission statement (unchanged since 2010, and the same in 2014): "Step one, solve intelligence. Step two, use it to solve everything else." — Source: Antoine Buteau, "Lessons from Demis Hassabis"
  • Interview with MIT Technology Review (December 2, 2014): "AI has enormous potential to be a wonderful thing for humanity. It will really accelerate progress in solving disease and in a whole range of areas where we're currently advancing only relatively slowly." — Source: MIT Technology Review, "Google's Intelligence Designer" (December 2, 2014)
  • Account of the Google acquisition (revealed years later): Google "willingly accepted DeepMind's ethical red lines," including a demand that the technology never be used for military purposes as a condition of the deal — Source: TIME, "DeepMind CEO Demis Hassabis Urges Caution on AI"

xAI

• Elon Musk (CEO, Tesla / SpaceX)
  • June 2014, CNBC's "Closing Bell": "I like to keep an eye on what's going on with artificial intelligence. I think there's potentially a dangerous outcome there." "There have been movies about this, you know, like Terminator—there are some scary outcomes, and we should try to make sure the outcomes are good, not bad." — Source: CNBC, "Musk warns AI is 'more dangerous than nukes'" (August 2014)
  • August 3, 2014, post on Twitter (X): "Worth reading Superintelligence by Bostrom. We need to be super careful with AI. Potentially more dangerous than nukes." — Source: CBS News (August 4, 2014)
  • A follow-up post on Twitter the same day: "Hope we're not just the biological boot loader for digital superintelligence. Unfortunately, that is increasingly probable." And, on the subject of AI risk: "Also worth reading Our Final Invention by @jrbarrat." — Source: NBC News
  • October 24, 2014, MIT Aeronautics and Astronautics Centennial Symposium: "I think we should be very careful about artificial intelligence. If I had to guess at what our biggest existential threat is, it's probably that. So we need to be very careful with artificial intelligence." "With artificial intelligence we are summoning the demon. In all those stories where there's the guy with the pentagram and the holy water, he's sure he can control the demon. It doesn't work out." "There should probably be some regulatory oversight, at the national and international level, just to make sure that we don't do something very foolish." — Source: The Washington Post (October 24, 2014), TechCrunch

Mistral AI

• Arthur Mensch (student / researcher)
  • None of note

DeepSeek

• Liang Wenfeng  (High-FlyerlCapital)
  • None of note

Meta

• Mark Zuckerberg（Facebook CEO）
  • December 2013 (at the announcement of FAIR, which ramped up its activities in 2014): "The goal here is to use new approaches in AI to help make sense of all the content that people share so we can generate new insights about the world to answer people's questions. In the future, I think we'll be able to build services that are much more natural to interact with and can help solve many more problems than any existing technologies today." — Source: "Facebook has big plans for AI" (December 10, 2013)
• Alexandr Wang (high school student, then MIT)
  • None of note

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • None of note

Microsoft

• Satya Nadella（MicrosoftCEO）
  • February 4, 2014, email to all employees on his first day as CEO: "Our industry does not respect tradition — it only respects innovation." "Our job is to ensure that Microsoft thrives in a mobile and cloud-first world." "I believe over the next decade computing will become even more ubiquitous and intelligence will become ambient. The coevolution of software and new hardware form factors will intermediate and digitize many of the things we do and experience in business, life and our world. This will be made possible by an ever-growing network of connected devices, incredible computing capacity from the cloud, insights from big data, and intelligence from machine learning." — Source: Microsoft News Center, "Satya Nadella email to employees on first day as CEO"
  • March 27, 2014, at the Mobile First, Cloud First press conference: "This business is exciting because, in a sense, it doesn't respect the tradition of what we've done in the past. It's all about innovation and moving forward. And what defines us, and will define Microsoft going forward, is a mobile-first, cloud-first world." — Source: Microsoft Stories
  • July 10, 2014, second company-wide memo: "We will reinvent productivity for people who are swimming in a growing sea of devices, apps, data and social networks." — Source: Cloud Computing News, "Microsoft CEO Nadella's memo"

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • None of note

2015: Events, Positions, and Statements

Key Events

• January 2–5: FLI's "Future of AI" conference in Puerto Rico brings together some 80 participants, including Hassabis, Musk, Russell, and Bostrom
• February: DeepMind's "Atari DQN" paper appears in Nature, marking the starting point of deep reinforcement learning
• July: FLI (the Future of Life Institute) publishes an open letter calling for a ban on autonomous weapons, signed by more than 3,000 people including Hassabis, LeCun, and Horvitz.
• November 9: Google releases TensorFlow as open source
• December 11: OpenAI is founded with a $10 billion funding package (Altman, Musk, Brockman, Sutskever, and others)

Positions and Statements by Individual

OpenAI

• Sam Altman (Y Combinator CEO → OpenAI Co-Chair)
  • February 25, 2015, "Machine Intelligence, Part 1": "Development of superhuman machine intelligence (SMI) is probably the greatest threat to the continued existence of humanity. There are other threats that I think are more certain to happen (for example, an engineered virus with a long incubation period and a high mortality rate), but they are unlikely to wipe out every human in the universe the way SMI could." "SMI does not have to be the inherently evil sci-fi version to kill us all. A more probable scenario is that it simply doesn't care about us much either way, and wipes us out in the course of pursuing some other goal (most goals, if you think about them long enough, could make use of the resources humans currently rely on)." "Evolution will keep moving forward, and if humans are no longer the fittest species, we may go extinct." "Nick Bostrom's excellent book Superintelligence is the best thing I've read on this topic. It is well worth a read." — Source: Sam Altman's blog – Machine Intelligence, Part 1; LessWrong link post
  • February 26, 2015, "Machine Intelligence, Part 2": "The US government, and all other governments, should regulate the development of SMI. In an ideal world, regulation would slow down the bad actors and speed up the good ones. What happens with the first SMI to be developed seems likely to matter a great deal." "It's probably impossible to stop the development of SMI. The best-case scenario is that we get enough of a head start to develop it carefully, with safety in mind." — Source: Sam Altman's blog – Machine Intelligence, Part 2; Data Economy
  • December 11, 2015, OpenAI founding statement (as Co-Chair): "Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return." — Source: TechCrunch – OpenAI announcement
• Greg Brockman (Stripe CTO → OpenAI CTO)
  • No notable statements

Human

• Dario Amodei (Baidu Silicon Valley AI Lab)
  • No notable statements
• Daniela Amodei (Risk Management at Stripe)
  • No notable statements

Google

• Sundar Pichai (Google Product Chief → Google CEO)
  • August 2015, blog post upon being named Google CEO: "It is clear to us and our board that it is time to simplify our management structure." (from the announcement issued jointly with Larry Page)
  • 2015, Forbes interview: declared that machine intelligence would sit at the core of Google's future strategy — Source: Google – CEO Sundar Pichai on Google Assistant
• Demis Hassabis (DeepMind CEO)
  • 2015, commenting on the publication of the Atari DQN paper: "I want to be clear about what we mean by 'general' here: we used the same system to learn 49 different Atari games. There was no programming for any individual game—we simply gave it the goal of 'play well.' So in that sense, this is a general system. But of course, compared with human capabilities, it is still very much a narrow AI." — Source: [DeepMind official statement, Nature interview]
  • 2015, responding to the AI-risk concerns raised by Stephen Hawking and Musk: "This technology has the potential to bring immense benefits, but it is also a dual-use technology. So if it falls into the hands of someone with bad intentions, it could be misused." — Source: Aifi – Demis Hassabis quotes
  • July 2015, signed the FLI open letter calling for a ban on autonomous weapons — Source: FLI 2015 Annual Review

xAI

• Elon Musk (CEO of Tesla / SpaceX → OpenAI Co-Chair)
  • January 15, 2015, announcing a $10M donation to FLI: "to fund research on artificial intelligence safety" (statement of the donation's purpose) — Sources: Immortal.org; FLI 2015 Annual Review
  • January 2015, signed the FLI open letter "Research Priorities for Robust and Beneficial AI": "Our AI systems must do what we want them to do." (the letter's key message, drafted by Stuart Russell) — Source: FLI open letter
  • May 2015, Vanity Fair interview (by Nick Bilton): "Musk believes it is better to build super-AI first and spread the technology across the world than to allow the algorithms to be hidden away and concentrated in the hands of a tech or government elite—even when that tech elite includes his own friends, such as Google founders Larry Page and Sergey Brin." — Source: Vanity Fair (via Quora)
  • July 2015, signed the FLI open letter calling for a ban on autonomous weapons
  •  December 11, 2015, media statement at OpenAI's founding: "It's hard to fathom how much human-level AI could benefit society, and it's equally hard to imagine how much it could damage society if built or used incorrectly." — Source: TechCrunch

Mistral AI

• Arthur Mensch (student / researcher)
  • No notable statements

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • No notable statements

Meta

• Mark Zuckerberg（Facebook CEO）
  • June 2015, Facebook Q&A (answering a question about AI): "One of our goals for the next five to ten years is to get better than human level at all of the primary human senses—vision, hearing, language, and general cognition. Taking in sensory input and understanding what it means is enormously valuable." — Source: Facebook Q&A
  • 2015, responding to Musk's claim that "AI is more dangerous than nuclear weapons" (a view he would go on to reject again in 2016 and 2017): "I think AI is going to bring a lot of good things—identifying and curing diseases, driving cars safely, keeping our communities safe, and much more."
• Alexandr Wang (high school student → MIT)
  • No notable statements

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • November 10, 2015, statement in the press release announcing the Tesla M40 GPU: "Deep learning is the most exciting development to come out of computing in the past five years, and it will be enormously important to everything we do." "GPU-accelerated computing is the foundation of deep learning." — Source: NVIDIA official press release (November 2015)
  • 2015, interviews with The Wall Street Journal and others: "This is a fundamentally new type of computing, and it will reshape the entire computing industry." (on GPUs and deep learning)

Microsoft

• Satya Nadella（MicrosoftCEO）
  • 2015, on his philosophy of AI and machine learning: "Ultimately, it's not going to be man versus machine—it's going to be man with machine." — Source: BrainyQuote – Satya Nadella

• Mustafa Suleyman (DeepMind Chief Product Officer)
  • July 2015, signed the FLI open letter calling for a ban on autonomous weapons in his capacity as DeepMind co-founder — Source: FLI 2015 Annual Review
  • 2015, commenting on the launch of DeepMind Health: "Our goal is to support clinicians by providing the technical expertise needed to build tools that help them deliver the best possible care to their patients." — Source: DeepMind official

2016: Events, Claims, and Statements

Key Events

• March 9–15: AlphaGo defeats Lee Sedol 4–1, watched by 200 million people worldwide—a historic moment for the AI industry
• April 27: OpenAI Gym released, an open platform for reinforcement learning research
• June: Google unveils the TPU (Tensor Processing Unit) at Google I/O
• September 28: Partnership on AI founded (Microsoft, Google, Facebook, Amazon, IBM, DeepMind)
• November: Google Translate switches to Neural Machine Translation (GNMT), dramatically improving translation quality

Claims and Statements by Each Figure

OpenAI

• Sam Altman (Y Combinator CEO / OpenAI Co-Chair)
  • October 2016, The New Yorker, "Sam Altman's Manifest Destiny" (Tad Friend): "I see no reason at all to believe that in about thirteen years there won't be hardware capable of replicating my brain. Yes, certain things about being human remain—creativity, that flash of inspiration from nowhere, the ability to feel joy and sadness at the same time—but computers will have their own desires and goal systems." "When I realized that intelligence could be simulated, I let go of the idea that humans are unique, and it wasn't as traumatic as I'd expected." "There are certain advantages to being a machine. We humans are limited by our input-output rate—we can only learn about two bits per second—so a vast amount of information is lost. To a machine, we must look like slowed-down whale songs." — Sources: The New Yorker – Sam Altman's Manifest Destiny (October 10, 2016); Daring Fireball – Friend's 2016 profile
  • 2016, on Y Combinator's Hal9000: "The thing about neural networks is that you have no idea what they're doing, and they can't tell you either." — Sources: The New Yorker – Sam Altman's Manifest Destiny (Oct 10, 2016); Daring Fireball – Friend's 2016 Profile
  • 2016, on AI "babies": "What people tend to forget about human babies is that it takes them years to learn anything interesting. If AI researchers were developing an algorithm and happened to stumble across the one for a human baby, they'd get bored watching it, conclude it wasn't working, and shut it down." — Source: The New Yorker
  • 2016, a Rickover quote posted in an OpenAI conference room: "The great end of life is not knowledge but action. I believe it is the duty of each of us to act as if the fate of the world depended on him. We must live for the future, not for our own comfort or success." 
• Greg Brockman (OpenAI Chief Technology Officer)
  • May 2016, when Dario Amodei paid a visit to Sam Altman and Greg Brockman (as recounted years later in The New Yorker): "Our goal right now… is to do the best thing there is to do." — Source: MIT Technology Review – OpenAI's bid to save the world
  • 2016, blog post announcing OpenAI Gym: "OpenAI Gym is a toolkit for developing and comparing reinforcement learning algorithms." "We're releasing the public beta of OpenAI Gym, a toolkit that makes it easy to develop and compare reinforcement learning agents." — Source: OpenAI blog post (2016)
  • 2016, on his relationship with Ilya Sutskever (recalled in later years): "In August 2015, OpenAI was nothing more than an idea discussed over dinner among Elon Musk, Sam Altman, Ilya Sutskever, myself, and a few others." "Ilya is best described as an artist who expresses himself through machine learning." — Source: Greg Brockman's blog - #define CTO OpenAI

Human

• Dario Amodei (Baidu→Google Brain)
  • June 2016, the paper "Concrete Problems in AI Safety" (Amodei as lead author, with Olah, Steinhardt, Christiano, Schulman, and Mané): "We discuss five practical research problems related to accident risk, categorized according to whether the problem originates from having the wrong objective function ('avoiding side effects' and 'avoiding reward hacking'), an objective function that is too expensive to evaluate frequently ('scalable oversight'), or undesirable behavior during the learning process ('safe exploration' and 'distributional shift')." — Source: arXiv:1606.06565 - Concrete Problems in AI Safety
  • 2016, research direction at Baidu when the Deep Speech 2 paper was published: led the approach in which "Deep Speech 2 recognized two vastly different languages, English and Mandarin, with accuracy on par with human workers" — Source: the Deep Speech 2 paper (arXiv:1512.02595)
• Daniela Amodei (risk management at Stripe)
  • Nothing notable

Google

• Sundar Pichai（Google CEO）
  • May 18, 2016, Google I/O 2016 keynote: "Looking at the future of computing, we see machine learning and artificial intelligence unlocking capabilities that were unthinkable just a few years ago." "This means that the power of software—the 'smarts'—matters more to hardware than ever before." "The last ten years have been about building a mobile-first world. The next ten years will see us shift to an AI-first world, one where computing becomes universally available—at home, at work, in the car, on the go—and where interacting with all of these surfaces becomes more natural, intuitive, and intelligent." "That's why we built the Google Assistant, so you can have a natural conversation with Google." — Source: YourStory - How Google moved from mobile-first to AI-first
  • 2016, internal executive meeting: "This is going to shape Google's future for the next decade or so." — Source: TIME - How Sundar Pichai pushed Google to the front of the AI race (2026)
  • April 28, 2016, Alphabet supplementary document: "That's why investing in machine learning and AI is a top priority for us. The implications of this win (AlphaGo versus Lee Sedol) are literally game-changing—and the ultimate winner is humanity. This is another important step toward creating artificial intelligence that can help us in everything from getting through our daily tasks and travels to eventually tackling even bigger challenges like climate change and cancer diagnosis." — Source: NBC News - Google's CEO points to the shift from a mobile world to an AI world
• Demis Hassabis（Google DeepMind CEO）
  • March 9, 2016, after AlphaGo won the first game (Twitter): "#AlphaGo WINS!!!! We landed it on the moon. So proud of the team!! Respect to the amazing Lee Sedol too." — Source: British Go Association - AlphaGo wins Game 1
  • March 10, 2016, after AlphaGo's Game 2 victory: "I think this is a testament to Lee Sedol's incredible skill. According to the commentators, AlphaGo played some truly surprising and beautiful moves, which was really wonderful to see." — Source: American Go E-Journal - AlphaGo wins again
  • March 13, 2016, on Twitter after Lee Sedol won Game 4: "Lee Sedol wins Game 4!!! Congratulations! He was too strong for us today, and #AlphaGo made a mistake it couldn't recover from." "It shows just how deep Go is. Lee Sedol's move 78 in Game 4 and #AlphaGo's move 37 in Game 2 will be discussed for a long time to come." — Source: IBTimes UK - Go world champion Lee Sedol claims his first win
  • March 2016, during preparations for the AlphaGo match, when researchers objected that "random moves should be avoided" (revealed in later years): "We are developing AI not to imitate human thinking, but to explore the unknown limits of intelligence." — Source: 36Kr - The 10th anniversary of the AlphaGo match
  • March 2016, summing up after the AlphaGo match: "We have landed on the moon." (attested in multiple sources) — Source: Seoul Economic Daily, UPI

xAI

• Elon Musk (CEO of Tesla / SpaceX → co-chair of OpenAI)
  • June 1, 2016, Code Conference 2016 (Recode): "The odds that we're all living in a simulation are extremely high." "In other words, there's only a one-in-billions chance that this is base reality." — Source: Code Conference 2016 transcripts
  • 2016, remarks at the Code Conference on the dangers of AI: "The best-case scenario is that we end up with a benign overlord... something like an ant colony or a zoo, from humanity's point of view." "With artificial intelligence, we are summoning the demon." (continuing his 2014 remark)
  •  April 2016, internal email to a Neuralink colleague (later disclosed in court): "DeepMind is growing at a very fast pace. I'm concerned that OpenAI isn't growing fast enough to catch up. In hindsight, setting it up as a nonprofit may have been the wrong call." — Source: CNN Business - Key takeaways from the Musk v. Altman trial
  • 2016, concerns about DeepMind: "DeepMind is causing me extreme mental stress. If they win, their 'one mind to rule the world' philosophy is going to be really bad news." "I'm concerned that Demis Hassabis could establish an AGI dictatorship." — Source: Transformer News - The OpenAI emails (a 2017 email from Brockman and Sutskever referencing Musk's 2016 remarks)
  • August 2016, on receiving the NVIDIA DGX-1: "Awesome." (his verdict to Jensen Huang) — Source: Fortune - Jensen Huang and Elon Musk's DGX-1 delivery

Mistral AI

• Arthur Mensch (student / researcher)
  • Nothing notable

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • Nothing notable

Meta

• Mark Zuckerberg（Facebook CEO）
  • December 19, 2016, "Building Jarvis" Facebook post (wrap-up): "My goal was to learn about the state of artificial intelligence—where we're further along than people realize, and where we're still a long way off." "Working on these challenges, I've always learned more than I expected. This one gave me a much deeper understanding of the internal technology Facebook engineers get to use, as well as a thorough overview of home automation." "For assistants like Jarvis to be able to control everything in the home for more people, more devices need to be connected, and the industry needs to develop common APIs and standards for devices to talk to each other." "One thing that surprised me about communicating with Jarvis is that when I have the choice between speaking and texting, I text far more often than I would have expected." "As CEO of Facebook, I don't get much time to write code in our internal environment. I've never stopped coding, but these days I mostly work on personal projects like Jarvis." — Source: Fast Company - A home visit with Mark Zuckerberg and Jarvis; TIME - Mark Zuckerberg's AI butler
  • 2016, an implicit rebuttal to Musk's fears of AI-driven extinction: "AI is going to unlock a huge number of positive things—identifying and curing diseases, driving cars safely, keeping our communities safe, and much more." — Source: multiple Facebook Live Q&A sessions
• Alexandr Wang (high school student → MIT)
  • Nothing notable

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • April 5, 2016, press release announcing the DGX-1: "The DGX-1 is easy to deploy and was created for one purpose: to unlock superhuman capabilities and apply them to problems that were once unsolvable." — Source: Fortune - Jensen Huang, Elon Musk, and OpenAI: hand-delivering the DGX-1
  •  August 2016, the message he wrote when hand-delivering the DGX-1 to OpenAI: "To Elon Musk and the OpenAI team! To the future of computing and humanity! I present you the world's first DGX-1!" — Source: Fortune - DGX-1 hand delivery
  • 2016, press announcement of the DGX-1 delivery to OpenAI: "I thought it was incredibly fitting that the world's first supercomputer dedicated to artificial intelligence would go to a research lab dedicated to open artificial intelligence." — Source: Fortune
  • 2016, an April 2016 email to Musk (later disclosed in the 2024 court case). Replying to Musk's email saying "OpenAI would like to purchase a DGX-1; it is unrelated to Tesla and is funded by Musk and a very small number of others": "This is the first supercomputer that sells itself on the web! Demand is coming in from all over the world." — Source: Videocardz
  • 2016, as he later revealed on the Joe Rogan Podcast: "When we announced the DGX-1, nobody in the world wanted one. We didn't get a single order. Nobody wanted to buy it, nobody wanted anything to do with it—except Elon. He said, 'Actually, there's a company that really needs this technology.' I thought, 'Wow, this is my first customer.'" — Source: Wccftech - Nobody but Elon Musk wanted NVIDIA's first AI supercomputer

Microsoft

• Satya Nadella（MicrosoftCEO）
  • March 30, 2016, Microsoft Build 2016 keynote: "Bots are the new apps. People-to-people conversations, people-to-digital-assistant conversations, people-to-bot conversations, and even digital-assistant-to-bot conversations. That's the world you're going to see in the years to come." — Sources: KSDK - Microsoft CEO: "Bots are the new apps"; InfoWorld - Satya Nadella bets the future on AI
  • June 28, 2016, Slate, "The Partnership of the Future": "Ultimately, humans and machines will work together—not against one another. Computers may win at games, but imagine what's possible when human and machine work together to solve society's greatest challenges like beating disease, ignorance, and poverty." "Now is the time for greater coordination and collaboration on AI." "The most productive debate isn't one of good versus evil: The debate should be about the values instilled in the people and institutions creating this technology." "The beauty of machines and humans working in tandem gets lost in the discussion about whether AI is a good thing or a bad thing. Our perception of AI seems trapped somewhere between the haunting voice of HAL in 2001: A Space Odyssey and the friendlier voices of today's personal digital assistants—Cortana, Siri, and Alexa." "AI must be designed to assist humanity… AI must be transparent… AI must maximize efficiencies without destroying the dignity of people… AI must be designed for intelligent privacy… AI must have algorithmic accountability… AI must guard against bias." (six AI principles) — Source: Slate - Microsoft CEO Satya Nadella: Humans and AI can work together
  •  September 26, 2016, Microsoft Ignite 2016: "We are not pursuing AI to beat humans at games. We want to use AI tools to empower people to build their own solutions." "Microsoft is positioning itself as the company building the Gutenberg press of the AI era." (the reporter's words) — Sources: ITProToday - Microsoft Ignite Satya Nadella AI; SD Times

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • February 24, 2016, statement at the launch of DeepMind Health: "At DeepMind Health, we use cutting-edge machine learning and systems neuroscience to build powerful general-purpose learning algorithms—the same ones that have helped us tackle some of the long-standing challenges in AI." "We believe AI and machine learning can revolutionize healthcare and help clinicians spend more time with their patients." — Sources: DeepMind official statement; multiple secondary sources
  • 2016, Financial Times interview on DeepMind Health: "We are working closely with frontline clinicians to make sure this technology is genuinely useful to them and focused on solving real problems."

2017: Events, Positions, and Statements

Key Events

• January 5–8: Asilomar conference on AI principles (organized by FLI) adopts 23 principles for AI development
• May: AlphaGo Master defeats Ke Jie, the world's strongest player, 3–0 (Future of Go Summit)
• June 12: The paper "Attention Is All You Need" is published (Google), giving birth to the Transformer architecture
• October: DeepMind announces AlphaGo Zero, which learns entirely through self-play without any human game records and beats AlphaGo Master 100–0
• December: DeepMind announces AlphaZero, surpassing humans at Go, chess, and shogi with a single unified architecture

Positions and Statements by Individual

OpenAI

• Sam Altman (CEO, Y Combinator / Co-Chair, OpenAI)
  • December 2017, "The Merge" (Medium): "A popular topic in Silicon Valley is talking about what year humans and machines will merge (or, if not, what year humans will get surpassed by rapidly improving AI or a genetically enhanced species)." "Everyone is talking about the merge, the singularity—the moment when humans and computers become the same thing." "Because so much of our self-worth is tied up in intelligence, we believe intelligence must be uniquely ours and that being only slightly higher than other animals on a continuum isn't good enough. AI will probably feel the same way and conclude that the difference between us and bonobos is hardly worth debating." — Sources: Medium - "The Merge" by Sam Altman; Y Combinator Podcast, Episode 100
  • March 2017, talk at Mosaic Ventures (London): "Taxing robots sounds nice, but if you think through the consequences it's foolish. That said, when it comes to taxing and redistributing wealth, we need to do far more, and that is going to be the future." "Technology equalizes capability and concentrates wealth. Total assets grow, but they end up in fewer hands." "We're not far from being able to eliminate poverty and provide universal healthcare. Enough food and a place to live will become basic human rights." "It will make people uncomfortable. It's nothing like socialism, but it's going to feel like it." — Source: Irish tech news - Y Combinator's Sam Altman speaks in London
  • March 1, 2017, TechCrunch Bullish: "There were some things I used to think were really important, and now there are new things that seem even more important." "Something I didn't expect has happened, and it's looking like it's going to be very bad. You have to be willing to reallocate your focus, your time, and your priorities." "Everyone is talking about the merge, the singularity, the moment when humans and computers become the same thing." — Source: TechCrunch - A conversation with Sam Altman on politics, startups, and technology
  • December 2017, Axios interview: "What's happening in America really frightens me. Now I understand a little better why." "It would be really good to have people in power who get ahead of [the economic changes], understand them, and think thoughtfully and compassionately about how they can benefit everyone." "The anti-Trump stuff is Sam's view, not YC's view." — Source: Axios - Sam Altman weighs in on tech's political moment
• Greg Brockman (Chief Technology Officer, OpenAI)
  • September 20, 2017, internal email to Altman and Musk, co-signed with Sutskever (made public years later in court proceedings): "Throughout this process, we haven't been able to fully trust your judgment." "Is AGI truly your primary motivation? How does it connect to your political goals?" "You've stated that you don't want to control the final AGI, but during this negotiation you've shown us that absolute control is extremely important to you." (addressed to Musk) "We are concerned that as the company makes genuine progress toward AGI, you will choose to retain absolute control of the company, despite your current intention." — Source: Transformer News - New OpenAI emails reveal a long history of distrust
  • August 2017, statement on OpenAI's Dota 2 1v1 victory: "This is a first step toward building AI systems that accomplish well-defined goals in messy, complicated situations involving real humans." — Source: [OpenAI blog - Dota 2 (August 2017)]

Human

• Dario Amodei (Google Brain → OpenAI)
  • April 2017, "Deep Reinforcement Learning from Human Preferences" (co-authored by Christiano, Leike, Brown, Martic, Legg, and Amodei): "For sophisticated reinforcement learning (RL) systems to interact usefully with real-world environments, we need to communicate complex goals to these systems." "In this work, we explore goals defined in terms of (non-expert) human preferences between pairs of trajectory segments. We show that this approach can effectively solve complex RL tasks without access to the reward function." — Source: arXiv:1706.03741 - Deep Reinforcement Learning from Human Preferences
  • 2017, on the direction of AI safety research: Following his collaboration with Google Brain, and around the time of his move to OpenAI, he systematized "RLHF" and "reward models," proposing the methods that would later become the foundational technology behind ChatGPT and GPT-4.
• Daniela Amodei (Risk Management, Stripe)
  • Nothing notable

Google

• Sundar Pichai（Google CEO）
  • May 17, 2017, Google I/O 2017 keynote: "Last year I spoke about the important shift in computing from a mobile-first to an AI-first world." "As you can see, we are beginning to understand images and video. Everything at Google started with our ability to understand text and web pages, so the fact that computers can understand images and video has profound implications for our core mission." "Voice and vision are becoming as important to computing as the keyboard." "We want to make machine learning accessible to hundreds of thousands of developers." — Source: CNBC - Google CEO Sundar Pichai on machine learning, big data, and IoT for everyone - Google I/O 2017
  • May 2017, on AutoML at Google I/O 2017: "This [AutoML] is one of the most promising ideas, because at its heart, it is neural networks designing other neural networks." — Source: Google I/O 2017 keynote
  • 2017, Alphabet letter following AlphaGo's victory over Ke Jie: "This year's progress in AI has been remarkable. AlphaGo's victory over the world's top-ranked Go player marks yet another milestone."
• Demis Hassabis（Google DeepMind CEO）
  • October 18, 2017, DeepMind official statement on the release of AlphaGo Zero: "It's amazing to see just how far AlphaGo has come in only two years. AlphaGo Zero is now the strongest version of our program and shows how much progress we can make even with less computing power and zero use of human data." "Ultimately we want to harness algorithmic breakthroughs like this to help solve all sorts of pressing real-world problems, like protein folding or designing new materials. If we can make the same progress on these problems that we have with AlphaGo, it has the potential to drive forward human understanding and positively impact all of our lives." — Source: Future of Life Institute - DeepMind's AlphaGo Zero
  •  October 2017, MIT Technology Review interview: "The most striking thing is that we don't need any human data anymore." — Source: MIT Technology Review – AlphaGo Zero shows machines can become superhuman without any help
  • October 2017, DeepMind blog post "AlphaGo Zero: Learning from scratch": "This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge. Instead, it is able to learn tabula rasa from the strongest player in the world: AlphaGo itself." 
  • May 2017, Future of Go Summit, following AlphaGo's victory over Ke Jie: "We've made a lot of improvements to AlphaGo since October and the games against Fan Hui. We'll have to see how they hold up against an opponent with the brilliance and creativity of Lee Sedol." (a pre-match preparation remark) — Source: CNBC – AlphaGo reaches the Lee Sedol AI milestone (from 2016, but cited in the context of AlphaGo's continued improvement)
  • 2017, background on the AlphaGo Zero paper (co-authored by David Silver): "By not using human data or human expertise in any fashion, we've actually removed the constraints of human knowledge. It's able to create knowledge itself from first principles." (David Silver) — Source: MIT Technology Review

xAI

• Elon Musk (CEO of Tesla / SpaceX → Co-chair of OpenAI)
  • July 15, 2017, National Governors Association (NGA) Summer Meeting, Rhode Island: "AI is a fundamental existential risk for human civilization, and I don't think people fully appreciate that." "I have exposure to the very cutting-edge AI, and I think people should be really concerned about it." "AI is a rare case where I think we need to be proactive in regulation instead of reactive, because I think by the time we are reactive in AI regulation, it's too late." "Robots will be able to do everything better than us... I'm not sure exactly what to do about this. This is really the scariest problem to me." "Until people see robots going down the street killing people, they don't know how to react, because it seems so ethereal." "Once there is awareness, people will be extremely afraid, as they should be." "Transport will be one of the first to go fully autonomous. But when I say everything—the robots will be able to do everything, bar nothing." — Sources: NPR – Elon Musk warns governors; CNBC – The 9 most surprising things Elon Musk said about robots and AI in 2017; Big Think
  •  July 25, 2017, Twitter reply to Zuckerberg's criticism of his "doomsday" warnings: "I've talked to Mark about this. His understanding of the subject is limited." — Source: CNN Money – Elon Musk says Mark Zuckerberg's understanding of AI is "limited"
  •  September 2017, on Twitter in response to Putin's remarks: "China, Russia, soon all countries with strong computer science. Competition for AI superiority at the national level is most likely the cause of WW3, in my opinion." "It begins..." (quote-tweeting Putin's remarks) — Source: CNAS – Putin and Musk are right
  • 2017, on AI regulation: "Nobody likes being regulated, but everything (cars, planes, food, drugs, etc.) that's a danger to the public is regulated. AI should be too." — Source: CNBC – The 9 most surprising things

Mistral AI

• Arthur Mensch (researcher)
  • Nothing of note

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • Nothing of note

Meta

• Mark Zuckerberg（Facebook CEO）
  • July 23, 2017, Facebook Live: "I have pretty strong opinions on this. I am optimistic. I'm basically an optimistic person. I think you can build things and the world gets better. But with AI especially, I am really optimistic." "I think people who are naysayers and try to drum up these doomsday scenarios—I just don't understand it. It's really negative, and in some ways I actually think it is pretty irresponsible." "In the next five to ten years, AI is going to deliver so many improvements in the quality of our lives. Just think about safety and health and keeping people safe. AI is already helping diagnose diseases better and match drugs to patients based on their condition so they can get better treatment." "One of the top causes of death for people is still car accidents, and if you can eliminate that with AI, that is going to be a dramatic improvement." "Whenever I hear people saying AI is going to hurt people in the future, I think: technology can generally always be used for good and bad, and you need to be careful about how you build it." — Source: CNBC – Mark Zuckerberg: Elon Musk's doomsday AI predictions are "pretty irresponsible"
  • 2017, on the practical application of AI: "Just think about safety and health and keeping people safe—AI is already helping us diagnose diseases more accurately and match the right medication to a patient's condition." — Source: CNBC – Mark Zuckerberg: Elon Musk's doomsday predictions
• Alexandr Wang（Scale AI CEO）
  • Nothing of note

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • May 10, 2017, GTC 2017 keynote (Tesla V100 announcement): "Volta is a giant leap forward—a new computing platform for AI." "Tesla V100 is the most advanced data center GPU ever built to accelerate AI, HPC, and graphics. A processor packed with 21 billion transistors, it is the most complex chip ever manufactured." — Source: NVIDIA official press release (May 10, 2017)
  • 2017, on NVIDIA's AI strategy: "Deep learning is the most exciting development in computing in the past five years, and it will bring even greater transformation over the next five." (continuing a message he had delivered consistently since 2015) — Source: NVIDIA official / GTC keynotes
  • 2017, on AI supercomputing: "GPU-accelerated computing is the foundation of deep learning. The combination of GPUs and AI will transform every industry."
  • 2017, on the democratization of AI: "We are creating the iPhone moment of AI computing." (the prototype of his later "iPhone moment of AI" remarks)

Microsoft

• Satya Nadella（MicrosoftCEO）
  • September 2017, in his book Hit Refresh (Harper Business): "Our industry does not respect tradition. It only respects innovation." "The way I read this is that for AI to scale, it will need to scale not just in capability but in trust." "AI must be designed to assist humanity." (reaffirming the six AI principles from his 2016 Slate essay) — Source: [Hit Refresh by Satya Nadella]
  • 2017, Microsoft Build 2017 keynote: "As an industry, we are on the cusp of a new era—the era of intelligence." "AI will become the runtime that shapes everything we do going forward." — Source: AI Business – Satya Nadella: AI will shape everything we do

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • March 2017, Digital Health interview: "We wouldn't have gotten into this if we didn't believe that in ten years' time we could fundamentally transform healthcare." "There are a ton of low-hanging-fruit opportunities where we can deliver great benefits to patients. So we're going to take our time and take advantage of those, and use them as a great opportunity to understand how to drive behaviour change in such a complex environment." — Source: Digital Health – What does Google DeepMind want from the NHS?
  • July 2017, response to the ICO's ruling that the Royal Free data-sharing arrangement was unlawful: "There's a fine line between finding exciting new ways to improve care and moving ahead of patients' expectations. We recognise that we fell short of this when we first began our work in healthcare, and we will continue to listen, learn, and improve." — Source: Google DeepMind – The Information Commissioner, the Royal Free, and what we've learned
  •  October 4, 2017, blog post announcing the founding of DeepMind Ethics & Society (DMES): "We believe AI can be of extraordinary benefit to the world, but only if held to the highest ethical standards." "Technology is not value neutral, and technologists must take responsibility for the ethical and social impact of their work." — Source: DeepMind official statement (October 2017)
  • 2017, general stance on AI ethics: "We need to examine the incentives in the technology industry and encourage business leaders to prioritize social responsibility alongside their fiduciary duties."

Events, positions, and statements in 2018

Key events

• February 20: Elon Musk steps down from the OpenAI board (citing a conflict of interest with Tesla's AI work)
• June 11: OpenAI announces GPT-1, the starting point for Transformer-based language models
• June: Google publishes its "AI Principles" (in the wake of internal backlash over Project Maven)
• October: Google announces BERT, a bidirectional Transformer that revolutionizes the field of NLP
• December: DeepMind's AlphaFold 1 places first at CASP13, the starting point for protein structure prediction

Positions and statements by each figure

OpenAI

• Sam Altman (CEO of Y Combinator / Co-chair of OpenAI)
  • February 20, 2018, OpenAI official blog (statement on Musk's departure): "As Tesla continues to become more focused on AI, this will eliminate a potential future conflict for Elon." "Elon will continue to advise and donate to the nonprofit." — Source: CNBC – Elon Musk is leaving the board of OpenAI
  • August 2018, TechCrunch Disrupt interview: "OpenAI's long-term goal is to build safe artificial general intelligence (AGI) that benefits humanity." "I believe AGI will be the most important technological development in human history." — Source: multiple secondary sources
  • 2018, Y Combinator Podcast, episode 100: Altman reaffirmed the ideas from his 2017 essay "The Merge," continuing to argue that the merging of humans and machines is already gradually underway. "I have a fixed amount of cognitive capacity I can use each day. I can spend it on anything. If I spend it on things that don't matter, I don't have the time to work on the things that really matter." — Source: Y Combinator – Sam Altman on choosing projects
  • 2018, on AI safety research: "We need to make sure that the systems we build are aligned with human values."
• Greg Brockman (Chief Technology Officer, OpenAI)
  • 2018, on his contribution to the GPT-1 paper, and a retrospective given years later: "In late August 2017, OpenAI's core members got together to discuss setting up a for-profit company to commercialize the technology and figure out how to raise the money needed to achieve artificial general intelligence (AGI)." "What triggered all of this was OpenAI's model beating top human players at the video game DOTA II. That event convinced everyone in the organization that, while compute was a critical resource for building powerful AI tools, raising funds as a nonprofit alone would not be enough." — Source: TechCrunch – How Elon Musk left OpenAI, according to Greg Brockman
  • 2018, internal records from the time of Musk's departure from OpenAI (made public years later in court): "There must never be a situation in which a single individual has complete and absolute control over OpenAI." — Source: TechCrunch – Brockman testimony, 2026
  • 2018, statement at the OpenAI Dota 2 announcement: "OpenAI Five is a step toward advanced AI systems that can handle the complexity and uncertainty of the real world."

Human

• Dario Amodei (Research Lead, OpenAI)
  • 2018, as part of OpenAI's collaborative research (with Christiano and others), in the paper "AI safety via debate" (May 2018): "In the long run, we want AI systems to be able to handle tasks that are extremely difficult for humans to evaluate." "Debate is a viable approach to AI alignment that can be combined with other approaches." — Source: [arXiv:1805.00899 - AI safety via debate (Irving, Christiano, Amodei)]
• Daniela Amodei (risk management at Stripe)
  • None in particular.

Google

• Sundar Pichai（Google CEO）
  • January 24, 2018, World Economic Forum in Davos: "AI is probably the most important thing humanity has ever worked on. I think of it as something more profound than electricity or fire." "Any time you work with technology, you need to learn to harness the benefits while minimizing the downsides." "I think a lot of things will turn out better than people expect. But the risks matter too." "The only way to solve some of these deeper problems is through a global, multilateral framework." — Sources: CNN Money – Sundar Pichai: AI is more profound than electricity or fire; World Economic Forum – Sundar Pichai's remarks at Davos; CNBC
  •  January 2018, Recode interview with Kara Swisher, when Swisher pushed back with "Fire's pretty useful, though": "Well, fire kills people too. Humanity learned how to harness fire for its own benefit, but we also had to overcome its downsides. So my point is that AI is extremely important, but we also have to be concerned about it." — Source: Fortune – Pichai: AI is more profound than fire
  • May 8, 2018, Google I/O 2018 keynote: "We're at an inflection point… Many of us grew up wondering when computers would become smart enough to do the tasks we thought only humans could do. We're in an exciting moment where that is becoming a reality." "Our push into machine learning and other technological innovations has to be careful and deliberate." "We feel a deep sense of responsibility to get this right." — Source: Arms Control Today – Google renounces AI development for weapons
  • June 7, 2018, blog post "AI at Google: our principles," published alongside Google's AI Principles: "AI is one of the most important things humanity is working on. It is a sophisticated technology with the power to change people's lives." "As a leader in AI, we feel a deep responsibility to get this right." "We will not pursue technologies that cause or are likely to cause overall harm." "We will not pursue weapons or other technologies whose principal purpose or implementation is to cause or directly facilitate injury to people." The seven principles: 1) be socially beneficial; 2) avoid creating or reinforcing unfair bias; 3) be built and tested for safety; 4) be accountable to people; 5) incorporate privacy design principles; 6) uphold high standards of scientific excellence; 7) be made available for uses that accord with these principles. — Source: Google Blog – AI at Google: our principles
• Demis Hassabis（Google DeepMind CEO）
  • December 2018, on Twitter after the CASP13 win (posted by Hassabis himself): "Proteins are essential to life. Predicting their 3D structure is a major unsolved problem in biology, with potentially huge implications for understanding disease and for drug discovery. Thrilled to announce that we won the CASP13 protein folding competition!" — Source: Synced – DeepMind's AlphaFold makes unprecedented progress in protein folding
  • December 2018, Guardian interview (after the CASP13 win): "We haven't solved the protein folding problem; this is just a first step." — Source: DeepMind AlphaFold
  •  December 7, 2018, DeepMind's official statement upon publication of the AlphaZero paper in Science: "AlphaZero starts from literally random play, with no knowledge beyond the basic rules of the game." "Using the same algorithm and hyperparameters, and starting with no knowledge other than the rules, it mastered three different complex games." — Source: DeepMind official blog (December 6, 2018), "AlphaZero: Shedding new light on the grand games of chess, shogi and Go"
  • 2018, DeepMind's consistent messaging on its AGI ambitions: "DeepMind's ultimate vision has always been to build AI and use it to accelerate the pace of scientific discovery, helping us deepen our knowledge of the world around us." — Source: DeepMind official statement
  • 2018, Hassabis's stance on AI safety: "This technology has the potential to bring immeasurable benefits, but it is also a dual-use technology." (A message he has repeated consistently since 2015.)

xAI

• Elon Musk (CEO of Tesla / SpaceX; departed OpenAI)
  •  February 2018, Musk's internal emails (made public years later in court), delivering a harsh verdict on OpenAI: "OpenAI is on a path to certain failure." (his words at the time of his departure, per Brockman's testimony) "I believe OpenAI has a 0% chance of being relevant to DeepMind or Google without a dramatic change in execution and resources." — Sources: OpenAI – OpenAI and Elon Musk; TechCrunch – How Elon Musk left OpenAI
  • February 2018, Tesla internal email (proposing to fold OpenAI into Tesla): "OpenAI should use Tesla as its cash cow. Tesla is the only path that could even stand a chance against Google. Even then, the odds of becoming a counterweight to Google are low. Not zero, though." — Source: OpenAI – OpenAI and Elon Musk
  • April 2018, HBO documentary "Do You Trust This Computer?": "AI doesn't have to be evil to destroy humanity. If AI has a goal and humanity just happens to be in the way, it will destroy humanity as a matter of course, without even thinking about it. No hard feelings." "The least scary future I can think of is one in which AI has at least been democratized." — Source: HBO documentary
  • September 2018, at South by Southwest and in multiple interviews: "I think we should be very careful about AI." (a continuation of his earlier remarks) "AI is more dangerous than nuclear weapons."

Mistral AI

• Arthur Mensch (researcher, INRIA / DeepMind)
  • None in particular.

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • None in particular.

Meta

• Mark Zuckerberg（Facebook CEO）
  • April 10, 2018, testimony before the US Senate (opening statement): "It was my mistake, and I'm sorry. I started Facebook, I run it, and I'm responsible for what happens here." "It's clear that we didn't do enough to prevent these tools from being used for harm. That goes for fake news, foreign interference in elections, and hate speech, as well as developers and data privacy." — Sources: Wikipedia – Facebook–Cambridge Analytica data scandal; The Washington Post – hearing transcript
  • April 10–11, 2018, testimony before the US Congress (on AI): "Over the long term, building AI tools is going to be the scalable way to identify and root out most of this harmful content." "We're using AI more proactively to identify hate speech, content that incites violence, and harmful misinformation." "I'm optimistic that over a five-to-ten-year period we'll have AI tools that can get into the nuances, the linguistic nuances, of different types of content and flag things more accurately for our systems." — Sources: The Washington Post – Zuckerberg Senate hearing transcript; EFF – Despite what Zuckerberg's testimony may imply, AI cannot save us
  • May 2018, around the F8 conference: "Building AI is the most powerful tool we have for supporting our community." "AI is one of the most important technologies humanity is working on." (echoing Pichai's messaging)
• Alexandr Wang（Scale AI CEO）
  • None in particular.

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • September 13, 2018, NVIDIA GTC Japan keynote: "There's no question that AI powered by deep learning is being deployed all over the world, and we're seeing incredible growth in Japan as well." "The number of applications that leverage deep learning is growing exponentially, and hyperscale data centers can't just do one thing—they have to do everything." "As a result, usefulness and utilization go way up. If every node can run any model at the same time, the utilization of this server is maximized." — Source: NVIDIA Blog – GTC Japan 2018 keynote
  • September 2018, at the launch of the Tesla T4 GPU: "T4 is a universal inference accelerator." "With T4, we're bringing the power of Turing to every data center, every server room, and every edge device."
  • March 2018, at the DGX-2 announcement at GTC 2018: "DGX-2 is the world's first two-petaflop system." "Two petaflops of compute in a $399,000 box. Two years ago, that would have cost $500 million."
  • August 2018, at the launch of the Turing GPU: "Turing represents the biggest leap in computer graphics in more than a decade." "Real-time ray tracing has been the holy grail of our industry. Turing makes it possible."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • May 7, 2018, Microsoft Build 2018 keynote: "Privacy is a human right. All of us have to treat privacy as a human right." "GDPR is a fantastic start, and it sets a standard for how people around the world should think about privacy." "I think the thing we need to work on in 2018 is having more dialogue about the ethics and principles that can drive the engine of AI." "Artificial intelligence should be about what computers *should* do." "We need to give designers and developers the tools not just to build great AI, but to make the right choices. This should be a first-class engineering requirement." — Source: InfoQ - CEO Satya Nadella Delivers Day One Keynote at Microsoft Build 2018
  • May 22, 2018, "Leading AI Transformation" event in London: "At the core of the experiences [and applications] we create, and of the platform itself, is arguably AI. AI is the runtime that will shape everything we do going forward in terms of advances in applications and platforms." "We are at a stage where we make choices based on the fact that technological progress doesn't simply happen—it happens because humans make design choices. And those design choices need to be grounded in principles and ethics." "Unfortunately, the corpus of human data is riddled with bias, so we need to invest in tools that can remove bias when modeling language from that corpus." — Source: AI Business - Satya Nadella: AI Will Shape Everything We Do
  •  September 2018, Microsoft Ignite keynote: "If AI is trained for one purpose and then used for another, that is unethical. We need to address the unintended consequences of technology. That is the responsibility of all of us." — Source: ComputerWeekly - Microsoft CEO urges industry to focus on ethics and privacy
  • 2018, Microsoft's approach to responsible AI: "AI must be designed for humans. AI must augment human capabilities and experiences." (Continuing the consistent message he has delivered since 2016)

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • 2018, DeepMind's healthcare work: DeepMind's joint research with Moorfields Eye Hospital was published in Nature Medicine in August 2018, unveiling an AI system that diagnoses 50 eye diseases with accuracy on par with ophthalmologists — Source: Official DeepMind statement (August 2018)
  • August 2018, commenting on his inclusion in the TIME 100 list: "AI must be built by diverse teams, deployed transparently, and held to high ethical standards." — Source: [TIME 100 (2018)]
  • 2018, as co-chair of the Partnership on AI: "We need to think about the impact AI will have on society, on labor markets, and on power structures."
  • 2018, continued operation of DeepMind Ethics & Society (DMES): "Technology is not value-neutral, and technologists must take responsibility." (A consistent message since 2017)

Events, Claims, and Statements in 2019

Key Events

• February 14: OpenAI unveils GPT-2 (1.5B parameters; deemed "too dangerous to release" and rolled out in stages)
• March 11: OpenAI restructures into a capped-profit entity (OpenAI LP)
• July 22: Microsoft invests $1B in OpenAI and becomes its exclusive Azure partner
• December: DeepMind unveils MuZero, a reinforcement learning system that learns the rules of the game on its own

Claims and Statements by Individual Figures

OpenAI

• Sam Altman (CEO of Y Combinator / Co-chair of OpenAI)
  • March 11, 2019, official OpenAI LP announcement: "Over the next few years, we will need to invest billions of dollars in large-scale cloud computing, in attracting and retaining talented people, and in building AI supercomputers." "We want to increase our ability to raise capital while still serving our mission, and no pre-existing legal structure we know of strikes the right balance." "Our solution is to create OpenAI LP, a hybrid of a for-profit and a nonprofit—which we are calling a 'capped-profit' company." — Sources: OpenAI LP announcement; TechCrunch - OpenAI shifts from nonprofit to 'capped-profit' entity
  • July 22, 2019, Altman's comments on the announcement of Microsoft's $1 billion investment: "The creation of artificial general intelligence (AGI) will be the most important technological development in human history, with the potential to shape the trajectory of humanity." "Our mission is to ensure that AGI technology benefits all of humanity, and we're working with Microsoft to build the supercomputing foundation on which we'll build AGI." "We believe it's crucial that AGI is deployed safely and securely and that its economic benefits are widely distributed. We are excited about how deeply Microsoft shares this vision." — Sources: Microsoft Source - Announcement of partnership with OpenAI; Fortune - Microsoft Invests $1 Billion in Elon Musk-Founded OpenAI Partnership
  • 2019, explaining OpenAI LP to investors: "We are confident that if we succeed, we will create orders of magnitude more value than any existing company—and that virtually all of that value will be returned to the world." — Source: The OpenAI Files - Concerns about the restructuring
• Greg Brockman (Chief Technology Officer, OpenAI)
  • July 22, 2019, blog post on the announcement of Microsoft's investment: "We are partnering to develop a hardware and software platform within Microsoft Azure which will scale to AGI." "OpenAI is producing a sequence of increasingly powerful AI technologies, which requires a lot of capital for computational power." "The most obvious way to cover costs is to build a product, but that would mean changing our focus. Instead, we intend to license some of our pre-AGI technologies, with Microsoft becoming our preferred partner for commercializing them." "We believe that the creation of beneficial AGI will be the most important technological development in human history, with the potential to shape the trajectory of humanity." "A difficult technical path lies ahead of us, requiring a unified software engineering and AI research effort at massive computational scale—but technical success alone is not enough." — Sources: OpenAI - Microsoft invests in and partners with OpenAI; Visual Studio Magazine
  • March 2019, explaining the OpenAI LP "capped-profit" structure: "The nonprofit's board retains full control and can only take actions that advance our mission. As spelled out in our charter, that mission is to ensure that AGI benefits all of humanity." — Source: The OpenAI Files
  • 2019, after OpenAI Five's victory: Brockman's official statement on OpenAI Five (which defeated Dota 2 world champions OG in April), emphasizing the importance of compute

Human

• Dario Amodei (Research Director, OpenAI)
  • 2019, TechCrunch's profile of OpenAI's Research Director: "OpenAI's research and technology are led by Ilya Sutskever (chief scientist) and Greg Brockman (CTO), with Sam supporting on the other aspects of management" (via remarks from Jack Clark, which also positioned Amodei as one of the research leads) — Source: TechCrunch - OpenAI shifts from nonprofit to 'capped-profit' entity
  • 2019, establishing the AI safety agenda: "OpenAI LP currently employs around 100 people organized into three main areas: capabilities (advancing what AI systems can do), safety (ensuring those systems are aligned with human values), and policy." — Source: OpenAI LP
  • 2019, the theoretical underpinning of the staged GPT-2 release strategy: Research by the Safety team under Amodei laid the groundwork for the later "Scaling Laws" paper (2020) and the GPT-3 report (2020)
• Daniela Amodei (Operations, OpenAI)
  • Nothing notable

Google

• Sundar Pichai（Google CEO→Alphabet CEO）
  • December 3, 2019, memo as CEO of Alphabet: "Today is a big day for Google and Alphabet." "Larry and Sergey built Alphabet with an organizational structure that lets each Alphabet company, including Google, operate independently, while focusing on long-term investments." "I will continue to be very focused on Google and the deep work we're doing to push the boundaries of computing and build a more helpful Google for everyone." — Source: [Google blog post (December 3, 2019)]
  • May 7, 2019, Google I/O 2019: "At its core, Google is an information company. And the next frontier is where AI and machine learning power more helpful experiences for people." "Information is at the center of everything we do." "Building a more helpful Google for everyone means making sure our products work for everyone, regardless of where they live, what language they speak, or what device they use." — Source: Google I/O 2019 keynote
  • October 23, 2019, CEO statement on achieving quantum supremacy: "This is a significant scientific milestone for our team as we work toward building a practical quantum computer."
• Demis Hassabis（Google DeepMind CEO）
  • January 25, 2019, statement at the unveiling of AlphaStar: "This is a major step forward for AI, with potential applications that extend far beyond gameplay." "StarCraft has been a grand challenge for AI research for the past decade. Mastering it requires not only sophisticated algorithms, but also the ability to plan over long horizons under imperfect information and across a vast action space." — Source: [DeepMind blog (Jan 25, 2019)]
  • 2019, on the AGI timeline (interviews with the FT and Nature): "AGI is still a long way off, but the progress we are making is remarkable."
  • December 2019, statement accompanying the MuZero announcement: "MuZero matches AlphaZero's performance in chess and Go without being told the rules, and surpasses it on Atari." "This is an important step toward general-purpose algorithms capable of solving real-world problems." — Source: [DeepMind blog (Dec 2019)]

xAI

• Elon Musk (CEO of Tesla / SpaceX; departed OpenAI)
  • August 29, 2019, World AI Conference Shanghai (WAIC 2019), in conversation with Jack Ma: "The biggest mistake that AI researchers make is assuming they're intelligent." "AI will be vastly smarter than the smartest human. The biggest mistake researchers make is thinking they're smart compared to AI." "We're already cyborgs (referring to mobile phones as an extension of human capability), but we're extremely slow compared to machines." "AI will make jobs kind of pointless. Probably the last job will be writing AI software, and then eventually AI will write its own software." "If you can't beat them, join them. That's what Neuralink is about. Can we go along for the ride with AI?" "I don't know, man, that's like famous last words." (in response to Jack Ma saying he wasn't afraid of AI) — Source: GeekWire - Billionaires Jack Ma and Elon Musk debate
  • July 2019, message at the Neuralink unveiling: "With a high-bandwidth brain-machine interface, I think we can actually go along for the ride." "This is going to sound pretty weird... but ultimately we can achieve a sort of symbiosis with artificial intelligence."
  • 2019, on Tesla's AI efforts: "Tesla is a real-world AI company that happens to make cars." "Full self-driving will be the single most important thing for Tesla."

Mistral AI

• Arthur Mensch (research scientist, DeepMind)
  • Nothing notable

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • Nothing notable

Meta

• Mark Zuckerberg（Facebook CEO）
  • May 2019, F8 developer conference: "AI has the potential to be one of the most transformative technologies of our lifetimes." "We're investing heavily in AI to help solve some of the world's biggest challenges."
  • October 2019, speech at Georgetown University: "I believe people should be given a voice. I don't think it's right for a tech company to censor politicians or the news in a democracy." "As we build new tools, we need to find ways to protect free expression." (an implicit nod to the limits of AI moderation)
• Alexandr Wang（Scale AI CEO）
  • August 2019, message accompanying the Series C funding announcement: "Data is the new oil of the AI economy." "Every AI application needs high-quality data, and we're building the infrastructure that makes it possible."
  • 2019, on being named to Forbes 30 Under 30: "AI is fundamentally transforming every industry, and high-quality data is the foundation of every AI breakthrough."
  • 2019, on expanding into government and defense: "AI for national security is one of the most important application areas we can contribute to." (Scale AI secured a contract with the U.S. Department of Defense)

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • March 18, 2019, GTC 2019 keynote (announcing the Mellanox acquisition): "Mellanox is a leading global provider of end-to-end smart interconnect solutions for data center servers and storage systems." "This is a key step in NVIDIA's strategy to become the leader in AI computing." — Source: NVIDIA blog - Jensen Huang's GTC 2019 keynote on the data center
  • 2019, on closing the Mellanox deal: "The emergence of AI and data science has reshaped data center network architecture. Data-center-scale workloads of the future will demand enormous interconnect bandwidth."
  • 2019, on the inference market: "Inference will be one of the biggest opportunities in computing over the next decade." (the strategy behind the Tesla T4's inference-dedicated design)

Microsoft

• Satya Nadella（MicrosoftCEO）
  • July 22, 2019, Nadella's comment on Microsoft's $1B investment in OpenAI: "AI is one of the most transformative technologies of our time and has the potential to help solve many of our world's most pressing challenges." "By bringing together OpenAI's breakthrough technology with new Azure AI supercomputing technologies, our ambition is to democratize AI—while always keeping AI safety front and center—so everyone can benefit." — Sources: Microsoft Source - partnership with OpenAI; GeekWire - Microsoft invests $1 billion in OpenAI
  • May 2019, Microsoft Build 2019: "Our mission is to empower every person and every organization on the planet to achieve more." "AI will be the runtime that shapes everything we do going forward." (continuing the consistent message from 2018)
  • September 2019, Microsoft Ignite: "As technology becomes ever more pervasive, the importance of trust cannot be overstated. Trust is the new currency."

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • August 2019, official statement (via a DeepMind spokesperson) before Suleyman went on administrative leave: "After 10 hectic years, Mustafa is taking some time out." — Source: CNBC - DeepMind co-founder Mustafa Suleyman
  • 2019 (from a later response to questions from the Wall Street Journal): "As a co-founder at DeepMind, I drove people too hard and at times my management style was not constructive. I accept these criticisms and apologize unequivocally to those who have been affected." — Source: CNBC - DeepMind co-founder placed under investigation
  • December 2019, statement on moving to Google: "After 10 years at DeepMind, I'm joining Google to begin a new chapter focused on AI policy and product management." Detailed remarks were limited, but the move was widely read within the industry as a "political retreat from DeepMind."
  • 2019 (recounted years later on a podcast with Reid Hoffman): "There was a period in 2017–2018 when several of my colleagues raised complaints about my management style. I really screwed up. I was very demanding and pretty relentless." "I think that created an environment where people had quite unrealistic expectations about what they should deliver and by when." "I still feel deeply sorry about the impact that had on people and the hurt they felt." — Source: CNBC - Why DeepMind co-founder Mustafa Suleyman left Google to become a venture capitalist

2020: Events, Claims, and Statements

Key events

• May 28: OpenAI unveils GPT-3 (175B parameters), ushering in a new era for AI language models
• June 11: OpenAI API released in beta
• November 30: DeepMind's AlphaFold 2 reaches human-level accuracy at CASP14, effectively solving protein structure prediction
• December: Dario and Daniela Amodei and others leave OpenAI, laying the groundwork for Anthropic

Claims and statements by each figure

OpenAI

• Sam Altman (Co-chair, OpenAI)
  • July 19, 2020, on Twitter (regarding the GPT-3 hype): "The GPT-3 hype is way too much. It's impressive (thanks for the nice compliments!) but it still has serious weaknesses and sometimes makes very silly mistakes." "AI is going to change the world, but GPT-3 is just a very early glimpse. We have a lot still to figure out." — Sources: Synced - As the GPT-3 model wows the world, OpenAI's CEO suggests it's "way too much" hype; Slator - What you need to know about GPT-3
  • July 22, 2020, Twitter replies on bias (responding to Anima Anandkumar and Jerome Pesenti): "We share your concern about bias and safety in language models, and it's a big part of why we're starting off with a beta and have a safety review before apps can go live." "We're not (yet) serving billions of users in production, and we want to learn from our own experience and others' before we do." "We completely agree that companies like ours need to be extremely careful about the negative impact we could have on the world." "Thanks again for the comments—and if you have other thoughts or lessons from how Facebook is approaching this, we'd love to hear them!" — Source: Synced - As the GPT-3 model wows the world
  • September 22, 2020, OpenAI's official statement on Microsoft's exclusive GPT-3 license: "OpenAI has agreed to license GPT-3 to Microsoft for their own products and services, in addition to continuing to offer GPT-3 and future models via the OpenAI API, as part of the multiyear partnership announced last year." "The deal has no impact on continued access to the GPT-3 model through OpenAI's API." — Source: OpenAI - OpenAI licenses GPT-3 technology to Microsoft
• Greg Brockman (CTO, OpenAI)
  • May 19, 2020, Microsoft Build 2020, Brockman's comment on the Azure AI supercomputer announcement: "This is the world's first AI supercomputer, built exclusively for OpenAI on Microsoft Azure." "The system would rank among the top five publicly disclosed supercomputers in the world." — Sources: Microsoft Build 2020 keynote; official OpenAI statement
  • June 2020, message accompanying the GPT-3 API release: "The OpenAI API today provides a general-purpose 'text in, text out' interface, allowing users to try it on virtually any English-language task." — Source: [OpenAI - GPT-3 API blog post]
  • 2020, on OpenAI's direction: "Compute is a critical resource for developing powerful AI tools." (reinforcing a message he had held consistently since 2017)

People

• Dario Amodei (OpenAI Research Lead → departed)
  • January 23, 2020, the "Scaling Laws for Neural Language Models" paper: "Language modeling performance improves smoothly and predictably as we appropriately scale up model size, data, and compute." "Performance has a power-law relationship with each of the three scale factors when not bottlenecked by the other two." — Source: [arXiv:2001.08361 - Scaling Laws for Neural Language Models (Kaplan, McCandlish, Henighan, Brown, ..., Amodei)]
  • May 2020, the GPT-3 paper "Language Models are Few-Shot Learners," on which Amodei was one of 31 authors: "We show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches." — Source: [arXiv:2005.14165 - GPT-3 paper]
  •  On his reasons for leaving in 2020 (reflecting in later years): "There was a group of us inside OpenAI who, after the development of GPT-2 and GPT-3, believed two things very strongly." "Some people say we left because we didn't like the Microsoft deal. That's wrong." "Trying to argue against someone else's vision is deeply unproductive." "Gather people you trust and go make your own vision happen." — Sources: Inc. - Anthropic CEO Dario Amodei says a "difference in vision" is why he left OpenAI; Fortune - Anthropic's CEO explains why he quit
  • 2020 (same period), on AI safety research: as one of the lead authors on both the "Scaling Laws" and "GPT-3" papers, he argued that "scale is the key" and "safety is the key" at the same time — the origin of the design philosophy that would later shape Claude.
• Daniela Amodei (Operations, OpenAI)
  • Nothing of note

Google

• Sundar Pichai（Alphabet CEO）
  • December 9, 2020, internal memo on the Timnit Gebru dismissal (obtained by Axios): "I've heard the reaction to Dr. Gebru's departure loud and clear: it seeded doubts and led some in our community to question their place at Google." "I am deeply sorry for that, and I accept the responsibility of working to restore your trust." "We need to accept responsibility for the fact that a prominent Black, female leader with immense talent left Google unhappily." "We need to assess the circumstances that led up to Dr. Gebru's departure, examining where we could have improved and led a more respectful process." — Sources: CNBC - Google CEO Sundar Pichai's memo on Timnit Gebru's exit; Axios - Sundar Pichai's memo on Timnit Gebru's departure; CNN Business - Google to investigate the exit of a prominent AI ethics researcher
  • November 30, 2020, on Twitter following AlphaFold 2's win at CASP14: "This is an incredible advance for AI and for science. AlphaFold finding a solution to one of biology's grand challenges is a huge step forward, and could have a major impact on how we understand disease and develop drugs." — Source: CNBC - DeepMind solves the "grand challenge" of protein folding with its AlphaFold AI
  • May 2020, Google I/O 2020 (a statement issued in place of the event, which was cancelled due to COVID-19): "We are focused on using technology to help people, especially in difficult times like this pandemic."
• Demis Hassabis（Google DeepMind CEO）
  • November 30, 2020, on Twitter following AlphaFold 2's win at CASP14: "Thrilled to announce our first major breakthrough in applying AI to a grand challenge in science. #AlphaFold has been validated as a solution to the protein folding problem, and we hope it will have a big impact on disease understanding and drug discovery." — Sources: BusinessToday - Biology's 50-year-old protein folding mystery solved; CNBC - DeepMind solves the protein folding problem
  • November 30, 2020, official DeepMind statement (Drug Target Review): "DeepMind's ultimate vision has always been to build AI and then use it to help further our knowledge about the world around us by accelerating the pace of scientific discovery." "For us, AlphaFold represents the first proof point of that thesis." "This advance is our first major breakthrough in a long-standing grand challenge in science, which we hope will have a big real-world impact on disease understanding and drug discovery." — Source: Drug Target Review - A breakthrough in AI-based protein structure and folding prediction
  • November 30, 2020, CNBC interview: "DeepMind's ultimate vision has always been to build artificial general intelligence (AGI) and use it to help solve some of the world's most pressing problems." — Source: CNBC - DeepMind solves the grand challenge of protein folding
  •  2020, remarks by CASP14 chair Professor John Moult (the field's assessment of Hassabis and AlphaFold 2): "DeepMind has made a huge leap forward. A 50-year-old grand challenge in computer science has been largely solved." "It will have a major impact on drug discovery in the near future." — Source: CNBC
  • December 2020, at the CASP14 conference presentation: Hassabis himself gave a 30-minute talk walking through "AlphaFold's ideas, architecture, and publication plans," underscoring DeepMind's commitment to transparency. — Source: DeepMind's AlphaFold timeline

xAI

• Elon Musk (CEO of Tesla / SpaceX; departed OpenAI)
  • July 2020, New York Times interview: "It's hard to predict the future of AI, but it wouldn't be surprising if we had full self-driving in perhaps five or six years." "My take on why very smart people miss AI is that they think computers will never be as smart as they are."
  •  August 28, 2020, at the Neuralink V0.9 demo: "It's like a Fitbit in your skull with tiny wires." "Broadly speaking, Neuralink is trying to solve very hard problems of the brain and spine with a chip." "With enough bandwidth, and combined with AI, we could give people superhuman cognition." — Source: Official Neuralink statement (August 2020)
  • December 2020, Axel Springer interview: "I think we're headed toward a situation where AI is vastly smarter than humans, and I think that time frame is less than five years from now." "The question remains whether AI will like us, or whether it will care about us." "That doesn't necessarily mean things will be bad. It just means they'll be out of our control."
  • 2020, continuing remarks on AI: "AI is far more dangerous than nuclear weapons." (repeating a line from 2018) "We're headed toward AI surpassing human intelligence within five years."

Mistral AI

• Arthur Mensch (Research Scientist, DeepMind)
  • Nothing of note

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • Nothing of note

Meta

• Mark Zuckerberg（Facebook CEO）
  • July 29, 2020, US congressional antitrust hearing: "Facebook has succeeded by being more innovative than other companies." "AI is essential to our large-scale content moderation efforts." — Source: US congressional hearing record
  • 2020, remarks at Facebook Connect (the company's VR/AR conference): "AI is indispensable to making the metaverse a reality. We're investing heavily in AI research."
  • 2020, on PyTorch: "PyTorch has become one of the most popular tools in AI research, used by researchers all over the world."
  • 2020, continuing his messaging on AI: "AI will bring immense benefits over the next decade." (a consistent message since 2017)
• Alexandr Wang（Scale AI CEO）
  • 2020, remarks on winning a $91M US Army contract: "AI for national security is one of the most important applications of our technology." "We're proud to support the Department of Defense in adopting advanced AI capabilities."
  • 2020, on government AI strategy: "The United States must lead in AI to maintain its technological and economic advantage."
  • 2020, on data infrastructure: "No matter how advanced an AI system is, its performance depends on the quality of the data it learns from."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • May 14, 2020, NVIDIA GTC 2020 keynote (A100 launch): "The age of AI is in full swing." "A100 is the biggest leap in NVIDIA GPU performance to date. Built on the new Ampere architecture, A100 is at the heart of every modern data center." "The NVIDIA A100 GPU delivers 20x the performance and is an end-to-end machine learning accelerator, from data analytics to training to inference." — Source: Analytics India Magazine - Jensen Huang's top quotes from Nvidia GTC 2020
  • September 13, 2020, NVIDIA blog post announcing the proposed acquisition of Arm: "AI is the most powerful technology force of our time." "Uniting NVIDIA and Arm will create a company fabulously positioned for the age of AI — a premier computing company." — Source: Official NVIDIA statement (September 13, 2020)
  • October 2020, NVIDIA GTC 2020 fall edition (DPU launch): "The data processing unit (DPU) has become an essential element of the modern, secure, accelerated data center, in which CPU, GPU and DPU are combined into a single computing unit."
  • 2020, NVIDIA's long-term vision: "AI is the most disruptive technology of our time."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • May 19, 2020, Microsoft Build 2020 keynote (announcing the Azure AI supercomputer built for OpenAI): "We've built one of the top five publicly disclosed supercomputers in the world, delivering new infrastructure that enables the training of extremely large AI models on Azure." "This will be the foundation for the next generation of AI breakthroughs." — Source: Microsoft Build 2020 keynote
  • September 22, 2020, statement on Microsoft securing an exclusive license to GPT-3: "We see this as an incredible opportunity to expand our Azure-powered AI platform, democratize AI technology, and enable new products, services, and experiences." — Source: Official Microsoft statement (September 22, 2020)
  • 2020, on responsible AI (multiple remarks): "AI must be designed for humans." (a message carried forward since 2016) "We have a responsibility as an industry to ensure that AI is built and deployed safely."
  • 2020, on COVID-19 and AI: "AI is helping in the fight against COVID-19, from accelerating vaccine research to improving how healthcare is delivered."
  • 2020, on AI's impact on society: "Trust is the new currency. As technology becomes ever more pervasive, the importance of trust cannot be overstated." (consistent with his message at Ignite 2019)

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • 2020, in his capacity as a board member of Wired and The Economist: "AI policy needs to be examined thoroughly at the highest levels of government and industry." (a consistent message)
  • 2020, as co-chair of the Partnership on AI: "We need to think about the impact AI will have on labor markets, power structures, and society."

2021: Events, Positions, and Statements

Key Events

• January 5: OpenAI unveils DALL-E and CLIP, marking the starting point of text-to-image generation
• February: Anthropic founded (by Dario and Daniela Amodei, Tom Brown, and others; Series A funding of $124 million)
• June 29: GitHub Copilot technical preview released (powered by OpenAI Codex)
• August 10: OpenAI Codex released
• October: Inflection AI founded (Mustafa Suleyman, Reid Hoffman, Karen Simonyan)

Positions and Statements by Individual

OpenAI

• Sam Altman（OpenAI CEO）
  • March 2021, "Moore's Law for Everything" (moores.samaltman.com): "My work at OpenAI reminds me every day about the magnitude of the socioeconomic change that is coming sooner than most people believe." "Software that can think and learn will do more and more of the work that people now do." "The technological revolution is unstoppable." "The magnitude of socioeconomic change will arrive sooner than most people think." "In as little as ten years, AI could generate enough wealth to pay every adult in the United States $13,500 a year." — Sources: CNBC – OpenAI CEO Sam Altman says AI could cover the cost of a basic income; Edward Arun Jones – Moore's Law for Everything: Sam Altman's AI utopia [moores.samaltman.com]
  • 2021, the core proposal of "Moore's Law for Everything": the "American Equity Fund" (a concept that would levy a 2.5% tax on all land and companies and distribute the proceeds as a UBI to every adult aged 18 and over). "Altman believes this 'American Equity Fund' would drive growth, reduce poverty, and build a fairer society by ensuring that everyone shares in the benefits of capitalism." — Source: Audible – Sam Altman's blog post "Moore's Law for Everything"
  • 2021, an early comment on DALL·E 2 (made in a later year): "It's the most fun thing we've built to play with... and there's a kind of delight to it that I haven't felt from technology in a while." — Source: Ankur's newsletter – OpenAI's DALL-E 2
  • 2021, his consistent message on AGI timelines: Under the theme of "society after AGI," he put forward concrete policy proposals including UBI, tax reform, and the American Equity Fund. This was the year he established his vision of a post-AI society, in parallel with the AGI development under way inside OpenAI.
• Greg Brockman (Chief Technology Officer, OpenAI)
  • June 29, 2021, statement at the release of the GitHub Copilot technical preview: "Copilot is built on a new algorithm called OpenAI Codex, a descendant of GPT-3." "Where GPT-3 generates English, OpenAI Codex generates code." — Sources: The Verge – GitHub and OpenAI launch an AI Copilot tool; InfoQ – GitHub previews Copilot, an OpenAI-powered coding assistant
  • 2021, on the design philosophy behind GitHub Copilot: "Codex was trained on terabytes of publicly available code pulled from GitHub, along with sample code written in English." — Source: InfoQ – GitHub previews Copilot
  • 2021, on the deepening partnership with Microsoft: Together with Altman, he led the commercialization of Codex, with their calendars "almost completely filled" with Microsoft meetings. "Microsoft planned to use Codex to power GitHub Copilot, a code-completion tool that would be one of the company's first commercial AI products." — Source: BlockBeats – Why OpenAI is trying to catch up with Claude Code

Anthropic

• Dario Amodei（AnthropicCEO）
  • 2021, mission statement at Anthropic's founding: "Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems." — Sources: Anthropic company page; Built In – What is Anthropic?
  • 2021, on why Anthropic was founded (explained in later years at the Future of Life Institute, in Fortune, and with Lex Fridman): "Within OpenAI, after the development of GPT-2 and GPT-3, there was a group of people who strongly believed two things." "The first was the idea that the more compute you pour into these models, the better and better they get, with almost no end in sight." "The second was that you need something beyond simply scaling the models up — things like alignment and safety. Just throwing compute at a model doesn't teach it those values." "And there was a set of people who believed both of these ideas. We trusted each other and wanted to work together. So, with those ideas in mind, we went off and started our own company." — Sources: Contrary Research – Anthropic business breakdown and founding story; Inc. – Anthropic CEO Dario Amodei says a "difference in vision" is why he left OpenAI
  •  2021, the real reason for leaving OpenAI (clarified in later years): "Some people say I left because I didn't like the terms of the deal with Microsoft, but that's wrong." "Trying to fight someone else's vision is deeply unproductive." "Gather people you trust and go make your own vision happen." — Source: Inc. – Anthropic CEO Dario Amodei on his departure from OpenAI
  • 2021, Anthropic's "Race to the Top" strategy: "If you can build a company people want to join, adopt practices people consider reasonable, and hold your position in the ecosystem, then others will imitate you." — Source: Dario Amodei, CEO of Anthropic
  • 2021, on Anthropic's founding team (Future of Life Institute): "It's a group of people who have known each other for many years and who have been thinking about, and working on, the AI safety debate for a long time." — Source: KITRUM – The inspiring story of Daniela Amodei
• Daniela Amodei (President, Anthropic)
  • 2021, on how Anthropic came together as an organization: Amid the second wave of the COVID-19 pandemic, every meeting was held over Zoom; eventually a group of 15–20 people began meeting for weekly lunches at Precita Park in San Francisco. "Anthropic was formed in the middle of the COVID-19 pandemic, holding its meetings entirely online over Zoom during the second wave."
  • 2021, on Anthropic's organizational philosophy: The name derives from the Greek for "human-centered, oriented toward humans," and the company established a distinctive philosophy of organizational management within the AI industry. Anthropic = related to humans. — Source: Oreate AI blog – The rise of Anthropic's founding team

Google

• Sundar Pichai（Alphabet CEO）
  • May 18, 2021, LaMDA unveiled for the first time at Google I/O 2021: "LaMDA is our latest research breakthrough, adding a new piece to one of the most tantalizing parts of the puzzle: conversation." "LaMDA was trained on dialogue. During its training, it picked up on several of the nuances that distinguish open-ended conversation from other forms of language." "LaMDA has had a great many of these conversations, and they often take unexpected turns." — Sources: Google I/O 2021 Keynote; Google AI Blog
  • 2021, commenting on the open-sourcing of AlphaFold 2: "This is a remarkable advance for science. By making AlphaFold 2 freely available to the scientific community, we can accelerate progress in drug discovery and many other fields."
  • December 2021, at the announcement of DeepMind's Gopher: "LaMDA, Gopher, and other large language models represent a groundbreaking advance in AI."
• Demis Hassabis（Google DeepMind CEO）
  • July 2021, at the publication of the AlphaFold 2 paper in Nature: "AlphaFold has been used to fold all 200 million proteins known to science." "We are making this freely available to the scientific community." — Sources: DeepMind AlphaFold; Timeline of AlphaFold
  •  December 8, 2021, DeepMind's official statement at the Gopher announcement: Today we are releasing three papers on language models that reflect this interdisciplinary approach. They include a detailed study of a 280-billion-parameter transformer language model called Gopher, a study of the ethical and social risks associated with large language models, and a paper investigating a new architecture with better training efficiency. — Sources: DeepMind blog – Language modelling at scale; Towards Data Science – With Gopher (280B), DeepMind has become the undisputed leader in language AI
  •  2021, the warning sounded by the Gopher ethics paper: "Language-generation tools such as ChatGPT and its predecessor GPT-3 could accelerate the spread of disinformation, make government censorship and surveillance easier, and perpetuate harmful stereotypes under a veneer of objectivity." — Source: TIME – DeepMind CEO Demis Hassabis urges caution on AI
  • 2021, on the scientific significance of AlphaFold 2: "AlphaFold has now been used to predict the structures of nearly all of the roughly 200 million proteins known to science."

xAI

• Elon Musk (CEO of Tesla / SpaceX; departed OpenAI)
  • August 19, 2021, Tesla AI Day: "Tesla is much more than an electric car company. Tesla is doing advanced AI work in hardware, at both the inference level and the training level." "We are, in effect, building real-world AI." "Our cars are essentially robots on wheels." "Today we're announcing the Tesla Bot. It's intended to be friendly, of course." — Source: Tesla AI Day 2021 keynote
  • 2021, on the Dojo training computer: "Dojo will be the most powerful AI training computer in the world." "It's designed to process video data for autonomous driving efficiently."
  • 2021, on the Optimus humanoid robot: "In the future, physical work will be a choice. If you want to do it, you can, but you won't need to." "Having a humanoid robot with intelligence is an extraordinary thing." — Source: Tesla AI Day 2021

Mistral AI

• Arthur Mensch (researcher at DeepMind)
  • No notable statements

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • No notable statements

Meta

• Mark Zuckerberg（Facebook CEO）
  • October 28, 2021, Facebook Connect 2021, announcing the company's rebrand to Meta: "Today we're seen as a social media company, but in our DNA we are a company that builds technology to connect people, and the metaverse is the next frontier, just like social networking was when we got started." "We believe the metaverse will be the successor to the mobile internet." "From now on, we're going to be metaverse-first, not Facebook-first." — Sources: Rev – Meta (Facebook) Connect 2021 Metaverse Event Transcript; Qz – Three years since Facebook became Meta
  • October 2021, on the vision for the metaverse: "From desktop to web to mobile. From text to photos to video. But this isn't the end of the line. The next platform and medium will be even more immersive—an embodied internet where you're in the experience, not just looking at it. We call this the metaverse." "And you'll be able to do almost anything you can imagine: get together with friends and family, work, learn, play, shop, create, as well as entirely new categories that don't really fit how we think about computers or phones today." — Source: Rev – Meta Connect 2021 transcript
  • 2021, on the projected scale of the metaverse: predicted that "within a decade, a billion people will be using the metaverse, generating hundreds of billions of dollars in commerce" (in later years this fell far short, and the company booked $70B in losses) — Source: Qz – Three years since Facebook became Meta
• Alexandr Wang（Scale AI CEO）
  • 2021, Forbes Magazine interview, as the world's youngest billionaire: "At Scale, we believe AI will be the most important technology humanity ever creates." "The biggest bottleneck in AI today isn't algorithms—it's data. We're building the infrastructure that makes AI possible."
  • 2021, at the announcement of Scale AI's Series E: "Every major AI breakthrough of the past decade has been enabled by large volumes of high-quality data." "We're proud to support the world's most ambitious AI projects, from self-driving cars to large language models to national defense."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • April 2021, NVIDIA GTC 2021 keynote: "AI is the most powerful technology force of our time." "The age of AI has begun. AI needs accelerated computing, and accelerated computing needs NVIDIA." "NVIDIA Omniverse is the future of metaverse-scale 3D simulation and collaboration."
  • April 2021, at the announcement of the Grace CPU: "Grace is a cutting-edge CPU designed for AI workloads. It removes the memory bottleneck that has constrained the training of large models." "This is the first time NVIDIA has built a CPU for AI."
  •  October 2021, at the announcement of Megatron-Turing NLG (jointly with Microsoft): "With 530 billion parameters, Megatron-Turing NLG is the world's largest and most powerful generative language model." "NVIDIA and Microsoft are committed to advancing the state of the art in natural language processing." — Sources: NVIDIA official statement; Microsoft Tech Community blog
  • 2021, on Omniverse: "Omniverse is the platform for building and operating metaverse applications." "The future of AI will be built in Omniverse."
  • 2021, on the growth outlook for the AI market: "AI will revolutionize every industry." "The AI revolution is only just beginning."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • May 2021, Microsoft Build 2021 keynote: "Every company is becoming a digital company—and a software company. AI is becoming the runtime that underpins every application." "Azure AI is the most comprehensive AI platform in the cloud." — Source: Microsoft Build 2021 keynote
  • June 2021, at the release of the GitHub Copilot technical preview: "GitHub Copilot is a major step forward in developer productivity, made possible through our partnership with OpenAI." "This is only the beginning of what AI can do for developers."
    November 2021, Microsoft Ignite 2021, announcing general availability of Azure OpenAI Service: "Azure OpenAI Service puts the world's most powerful AI models in the hands of developers and enterprises." "By delivering AI through the trusted Azure platform, we're democratizing AI." — Sources: Microsoft Ignite 2021; Microsoft official statement
  • 2021, on Microsoft's strategy: "Tech intensity—the combination of adopting technology and building the capability to use it—is the key driver of business success."
  • 2021, on the metaverse (following Meta Connect): "The metaverse will fundamentally change how we work and play. Microsoft is investing heavily in this space."

• Mustafa Suleyman (Chief Product Officer, DeepMind)
  • No notable statements

2022: Events, Claims and Statements

Key Events

• April 6: OpenAI announces DALL-E 2, generating photorealistic images
• May: Google announces PaLM (540B parameters)
• July: DeepMind releases AlphaFold predictions for the full proteome (200 million proteins)
• August 22: Stable Diffusion released (Stability AI, open-source image generation)
• November 30: OpenAI launches ChatGPT, reaching one million users in five days—the Big Bang of the AI industry
• December 15: Anthropic publishes its paper on Constitutional AI.

Claims and Statements by Individual

OpenAI

• Sam Altman（OpenAI CEO）
  • November 30, 2022, on Twitter at the launch of ChatGPT: "Today we launched ChatGPT. Try talking with it here: http://chat.openai.com" — Source: Sam Altman on X (November 30, 2022)
  •  December 2022, commenting after ChatGPT went viral: "I expected everything to be about an order of magnitude smaller, on every axis. The hype would probably have been an order of magnitude smaller too." — Source: Fortune – The inside story of ChatGPT
  • December 2022, stating that ChatGPT costs "single-digit cents" per interaction: replying to Elon Musk on Twitter, he revealed that "each ChatGPT chat costs single-digit cents." — Source: Fortune – The inside story of ChatGPT
  • November 2022, commenting on the release of DALL·E 2: "It's the most fun thing we've ever built to play with... and there's a kind of joy to it that I haven't felt from technology in a while." — Source: Ankur's newsletter – OpenAI's DALL-E 2
  • 2022, continuing remarks on the AGI timeline: "AI is going to change the world in ways we can't fully predict."
  • December 2022, on internal goals at OpenAI (recalled later in Bloomberg): "We had to cover the costs somehow. I believe we tested two prices, $20 and $42, and most people felt $42 was a bit too much." — Source: Bloomberg – interview with Sam Altman
• Greg Brockman (Chief Technology Officer, OpenAI)
  • November 29, 2022, internal remark the day before ChatGPT launched (as reported by the NY Times): "At most one tweet thread with over 5,000 likes." (his expectation for the ChatGPT launch) — Source: Business Insider – OpenAI's president didn't expect ChatGPT to be such a hit
  • 2022, looking back on the ChatGPT launch: "Honestly, I was in the camp of 'I don't know if this is going to work.'" — Source: Fortune – The inside story of ChatGPT
  • December 2022, on the "surprise" of ChatGPT going viral: "This was a genuine surprise." (together with Mira Murati) 
  • 2022, on how ChatGPT came to be a Hail Mary (as recounted later): "It really was a last-ditch move: we decided to pull ChatGPT out of the development pipeline and release it so that ordinary users could try it." — Source: Fortune – The inside story of ChatGPT

Anthropic

• Dario Amodei（AnthropicCEO）
  • December 15, 2022, statement accompanying the publication of the "Constitutional AI" paper: "As AI systems become more capable, we would like to enlist their help to supervise other AIs." "We experiment with methods for training a harmless AI assistant through self-improvement, without any human labels identifying harmful outputs." "The only human oversight is provided through a list of rules or principles, and so we refer to the method as 'Constitutional AI.'" — Source: Anthropic – Constitutional AI: Harmlessness from AI Feedback (arXiv:2212.08073)
  •  2022, on why the first version of Claude was held back (made public in later years): "In the summer of 2022, Anthropic finished training the first version of Claude, but did not release it right away, citing the need for further internal safety testing and a desire to avoid triggering a potentially dangerous race to build more powerful AI systems."
• Daniela Amodei (President, Anthropic)
  • No notable statements

Google

• Sundar Pichai（Alphabet CEO）
  • September 2022, Vox Media Code Conference (responding to the Lemoine episode): "LaMDA is not sentient in any sense." "I think we're still a long way off. We often end up in philosophical or metaphysical debates about what sentience is and what consciousness is." "Done right, AI technology will 'empower humanity.'" — Source: Fortune – Google CEO Sundar Pichai rejects claims of sentient AI
  • May 2022, Google I/O 2022: "At Google, we focus on building products that organize the world's information and make it universally accessible and useful."
  • December 2022, internal response following the release of ChatGPT (as reported by The New York Times and others): Declared a "Code Red," brought Brin and Page back in, accelerated development of Bard, and reorganized teams to respond to ChatGPT — Source: KaaShiv InfoTech – Sundar Pichai on AI delays
  • Later in 2022, reflecting on the Code Red: "There was no way we could have released it at that point." (Referring to LaMDA, citing toxicity issues and insufficient sophistication) — Source: KaaShiv InfoTech
• Demis Hassabis（Google DeepMind CEO）
  • March 2022, the Chinchilla paper (Hoffmann et al., co-authored by Hassabis): "We investigate the optimal model size and number of tokens for training a transformer language model under a given compute budget." "We find that current large language models are significantly undertrained." "For compute-optimal training, the model size and the number of training tokens should be scaled equally: for every doubling of model size, the number of training tokens should also be doubled." "Chinchilla uniformly and significantly outperforms Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B)." — Source: arXiv:2203.15556 – Training Compute-Optimal Large Language Models, NeurIPS 2022
  •  July 2022, on the release of AlphaFold predictions for entire proteomes: "AlphaFold has now been used to fold all 200 million proteins known to science." "We are making this freely available to the scientific community." — Source: DeepMind – AlphaFold

xAI

• Elon Musk (CEO, Tesla / SpaceX)
  • September 2022, Tesla AI Day 2022: "The Tesla Bot will be able to take on all of the most boring, dangerous, and repetitive tasks." "This is what we're producing this year. Ramping up to mass production will move far faster than this." — Source: Tesla AI Day 2022 keynote
  • December 2022, exchange with Altman on Twitter after the release of ChatGPT: Musk asked Altman about the "cost per ChatGPT chat"; Altman replied that it was in the single-digit cents 
  • December 2022, on Twitter after the release of ChatGPT: "ChatGPT is scary good. We are not far from dangerously strong AI." 
  • 2022, on the Twitter acquisition: "In the relentless pursuit of truth" (Twitter's new mission)

Mistral AI

• Arthur Mensch (Research Scientist, DeepMind)
  • Nothing of note

DeepSeek

• Liang Wenfeng (High-Flyer Capital)
  • Nothing of note

Meta

• Mark Zuckerberg（Meta CEO）
  • November 2022, announcing the largest layoffs in Meta's history: "This is a sad moment, and there's no way around that. To those who are leaving, I want to thank you again for everything you've put into this place." "I got this wrong, and I take responsibility for that."
  • 2022, consistent messaging on the metaverse strategy: "We are building the metaverse together, and we will get there." (A message repeated since 2021) "This is a long-term investment, and we believe in it deeply."
• Alexandr Wang（Scale AI CEO）
  • October 2022, TransformX 2022 conference (in conversation with Emad Mostaque): "We believe AI will be the most important technology humanity ever creates." "The biggest bottleneck in AI today isn't algorithms — it's data." — Source: Scale events – TransformX 2022 with Emad Mostaque
  • 2022, on the demand for RLHF following the release of ChatGPT: "Every breakthrough in AI is made possible by large volumes of high-quality data."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • March 22, 2022, NVIDIA GTC 2022 keynote (announcing the H100 Hopper): "AI has fundamentally changed what software can do and how it is produced. Companies revolutionizing their industries with AI understand the importance of AI infrastructure." "Our new DGX H100 systems will power enterprise AI factories to refine data into our most valuable resource: intelligence." — Source: TechRadar – Nvidia GTC 2022 keynote; NVIDIA Blog – AI factories, Hopper H100
  •  March 2022, on AI's "stunning" progress: "AI has made stunning progress, and the next wave is arriving." "Companies are processing and refining their data, developing AI software, and transforming themselves into intelligence manufacturers." — Source: NVIDIA Blog – AI factories, Hopper H100
  • March 2022, on the performance of the H100 Hopper: "H100 is the new engine of the world's AI infrastructure." "H100 delivers an order-of-magnitude leap in performance for large-scale AI and HPC compared with the previous-generation NVIDIA A100." — Source: NVIDIA Developer Blog – Hopper architecture
  • 2022, the Earth-2 initiative (AI-driven climate modeling): "Scientists predict that effectively simulating regional climate change will require a supercomputer a billion times larger than those available today." "NVIDIA is taking on this grand challenge with Earth-2, the world's first AI digital-twin supercomputer." — Source: NVIDIA Blog
  • 2022, on Omniverse: "Omniverse enables digital twins that can be used for creative collaboration and for digital replicas of factories." "This is the second wave of AI."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • 2022, on Microsoft's 2022 strategy: "Every company is becoming a software company, and every software company is becoming an AI company." "AI is the runtime that will power every application."
  • May 2022, Microsoft Build 2022: "AI is the most transformative technology of our time, and we are at the dawn of a new era of computing."
  • November 2022, Microsoft's immediate internal response to the release of ChatGPT: Accelerated plans to integrate the technology into Bing AI and ramped up development of Copilot for Office and Microsoft 365
  • 2022, on responsible AI (a consistent message): "AI must be designed for humans." "Trust is the new currency. As technology becomes pervasive, the importance of trust cannot be overstated."
  • December 2022, strategy in response to the ChatGPT shock: Internally decided on additional investment in OpenAI and integration of its technology into Bing AI, laying the groundwork for an announcement in January 2023 — Source: multiple secondary sources; NBC News – Microsoft invests billions in ChatGPT maker OpenAI

• Mustafa Suleyman (Inflection AI CEO)
  • March 8, 2022, CNBC interview announcing the co-founding of Inflection AI: "If you look at the history of computing, we've always had to reduce the complexity of our ideas in order to communicate them to machines." "Even when we write a search query, we simplify, abbreviate, and use shorthand so that the search engine can understand what we mean." "It feels like we're on the cusp of being able to generate language at close to human-level performance. That opens up an entirely new suite of things we can do in the product space." — Source: CNBC – Reid Hoffman co-founds his first new company since selling LinkedIn
  • March 2022, Inflection AI's mission: "An AI-first consumer products company"; "AI that helps humans 'talk' to computers" 
  •  January 2022, Reid Hoffman's comment on Suleyman joining Greylock Partners: Reid Hoffman: "I've known Mustafa personally for more than a decade. We first met in a pub in London. What stayed with me most was the stimulating, thought-provoking discussion we had about the power and potential of AI to help solve the most pressing challenges facing humanity."
  • 2022, on the Pi (Personal Intelligence) concept: "All of these are ways in which we simplify our ideas and reduce their complexity — and, in a sense, sacrifice some of our creativity and originality — just to get a machine to do something." "AI is one of the most transformative technologies of our time." (A message shared with Reid Hoffman) — Source: CNBC
  • 2022, on the timeline for reaching AGI: "Within five years, it will almost certainly be achievable." (On human-level language generation) — Source: CNBC

2023: Events, Claims, and Statements

Key events

• January 23: Microsoft invests an additional $10B in OpenAI
• March 14: OpenAI announces GPT-4; Anthropic releases the first version of Claude
• March 29: The FLI open letter calling for a six-month pause is published, signed by 33,000 people including Musk and Wozniak.
• May 30: CAIS publishes its "Statement on AI Risk," signed by Hinton, Bengio, Altman, Hassabis, and others.
• July 18: Meta releases Llama 2 (open source, licensed for commercial use)
• November 17–22: Sam Altman is ousted from OpenAI, then reinstated four days later amid turmoil

Claims and statements by each figure

OpenAI

• Sam Altman（OpenAI CEO）
  • May 16, 2023, before the U.S. Senate Judiciary Committee: "My name is Sam Altman, and I am the Chief Executive Officer of OpenAI. OpenAI is a company that researches, develops, and deploys artificial intelligence (AI), and has built AI tools including ChatGPT, Whisper, and DALL·E 2." "OpenAI was founded on the belief that safe and beneficial AI offers tremendous possibilities for humanity." "I appreciate the opportunity to speak about our experience in developing cutting-edge AI technology and researching AI safety, and about our interest in working with government to ensure that safe and beneficial AI tools are developed and made widely available." — Source: OpenAI – Testimony before the U.S. Senate
  •  May 2023, "Governance of Superintelligence," co-authored by Altman, Brockman and Sutskever: "Given the picture as we see it now, it's conceivable that within the next ten years, AI systems will exceed expert skill level in most domains." "Given the possibility of existential risk, we can't just be reactive." Proposed an international AI oversight body modeled on the IAEA.
  •  May 2023, signatory to the CAIS "Statement on AI Risk": "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." — Source: official CAIS statement
  •  November 17, 2023, OpenAI's official statement announcing Altman's removal: "The board no longer has confidence in his ability to continue leading OpenAI." "Mr. Altman was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities."
  •  November 22, 2023, on Twitter upon returning to OpenAI: "I love OpenAI, and everything I've done over the past few days has been in service of keeping this team and its mission together."
• Greg Brockman (Chief Technology Officer, OpenAI)
  • March 14, 2023, GPT-4 developer livestream: "GPT-4 is far more capable than 3.5, especially when it comes to handling more complex, nuanced instructions." "It's not perfect — but neither are you. GPT-4 is an amplifier that can take you to new heights." — Source: Jake Miller – OpenAI announces GPT-4
  •  May 2023, co-authored "Governance of Superintelligence" (with Altman and Sutskever): "Superintelligence could arrive within the next ten years and bring about a 'dramatically more prosperous future.'" "Given the possibility of existential risk, we can't just be reactive."
  •  November 17, 2023, reaction to Altman's ouster: announced on Twitter, "Based on today's news, I quit," resigning alongside Altman → headed to Microsoft 
  • November 22, 2023, on Twitter upon his return: "We are so back." (Returned to OpenAI together with Altman)

Human

• Dario Amodei（AnthropicCEO）
  • May 2023, statement on the $450 million Series C round: "We are thrilled that these leading investors and technology companies are supporting Anthropic's mission of safety-first AI research and products." "The systems we are building are designed to deliver reliable AI services that can have a positive impact on businesses and consumers, now and in the future." — Source: Anthropic – Anthropic Series C funding
  • May 2023, signatory to the CAIS "Statement on AI Risk": "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war."
  • April 2023, as reported by TechCrunch (from an Anthropic strategy document): "Claude-Next will be ten times more capable than today's most powerful AI." "These models could begin to automate large portions of the economy." "Companies that train the best models in 2025/26 will be too far ahead for anyone to catch up in subsequent cycles." — Source: TechCrunch – Anthropic's $5B, four-year plan to take on OpenAI
  • August 2023, on thinking through AI safety: "The mature way to think about these things is not to deny that there are costs, but to think about what the costs and benefits are. I think we've acted relatively responsibly, in the sense that we didn't cause the kind of major acceleration that happened at the end of last year and the beginning of this year." — Source: Anthropic website
  • 2023, on Constitutional AI: "We write a document we call a constitution. [...] We tell the model, 'Okay, act in accordance with the constitution.'" "If we have to trade one off against the other, I'd much rather Claude be boring than dangerous. [...] Ultimately, we'll get the best of both." — Source: Anthropic website
  •  November 2023, turned down the OpenAI board's approach about becoming CEO: "Amodei declined both offers." (Both the CEO role and a proposed OpenAI–Anthropic merger) 
• Daniela Amodei (President, Anthropic)
  • 2023, establishment of the Anthropic Long-Term Benefit Trust: Anthropic's official messaging describes it as establishing "a governance structure for AI safety research and responsible AI development." — Source: Anthropic company page
  • 2023, on Anthropic's period of rapid organizational growth: "They're a group of people who have known each other for years, who have thought about and worked on the AI safety debate for a long time." (Continued via FLI/KITRUM) — Source: KITRUM – The inspiring story of Daniela Amodei

Google

• Sundar Pichai（Alphabet CEO）
  • April 2023, announcing the merger of Google Brain and DeepMind: "To accelerate our progress, we are bringing together two of the leading research groups in the AI field: the Brain team from Google Research, and DeepMind. Their collective accomplishments in AI over the last decade span AlphaGo, Transformers, word2vec, WaveNet, AlphaFold, sequence-to-sequence models, distillation, deep reinforcement learning, and distributed systems and software frameworks like TensorFlow and JAX for expressing, training and deploying large-scale ML models." — Sources: official Google blog, CNBC
  • May 10, 2023, Google I/O 2023 keynote: "We are at an exciting inflection point. We have an opportunity to make AI even more helpful for people, for businesses, for communities, for everyone." "We are reimagining all of our core products, including Search."
  •  December 6, 2023, on the launch of Gemini 1.0: "Gemini is the result of large-scale collaborative efforts by teams across Google, including our colleagues at Google DeepMind." "This is a significant milestone for us, and an inflection point on our journey to make AI more helpful for everyone." — Source: official Google statement
  • 2023, interview with 60 Minutes: "AI is too important not to regulate." "Society needs to adapt." "This is a profound technology and it needs to be considered carefully."
• Demis Hassabis（Google DeepMind CEO）
  • May 2023, signatory to the CAIS "Statement on AI Risk": "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." — Source: Aify – Demis Hassabis quotes
  •  May 2023, interview with The Verge (on AGI timelines): "I think there are legitimate concerns, and we should be discussing and considering them decades before anything of real consequence or influence actually happens, so that we can prepare adequately in advance." — Source: Aify – Demis Hassabis quotes
  •  April 2023, on the merger of Google Brain and DeepMind: "This will help us build more capable systems more safely and responsibly."
  •  May 2023, The New York Times (on AI extinction risk): explaining his signing of the CAIS statement: "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." — Source: Aify
  • December 6, 2023, on the launch of Gemini 1.0: "Gemini is the first model to outperform human experts on MMLU (Massive Multitask Language Understanding)." "Gemini Ultra significantly advances the state of the art on 30 of the 32 academic benchmarks widely used in large language model (LLM) research and development."
  • 2023, interview with TIME: "I would advocate not moving fast and breaking things." "They don't realize they're holding dangerous material." (A critique of the attitude of some AI researchers) — Source: TIME – DeepMind CEO Demis Hassabis urges caution on AI

xAI

• Elon Musk（xAI CEO）
  • March 29, 2023, signatory to the FLI "six-month pause" letter: "AI labs are locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one — not even their creators — can understand, predict, or reliably control." "Pause Giant AI Experiments: An Open Letter" — Sources: official FLI statement, multiple secondary sources
  •  July 12, 2023, on Twitter, officially announcing xAI: "Announcing formation of @xAI to understand reality." — Source: CBS News – Elon Musk launches new AI company
  • July 2023, explaining xAI on Twitter Spaces: "I think a maximally curious AI that strives to understand the universe will be pro-humanity, from the standpoint that humanity is much more interesting than not-humanity." "Maximally curious, maximally truth-seeking." — Source: Fortune – Elon Musk's xAI
  •  July 2023, on the dangers of politically correct AI: "A politically correct AI would be 'extremely dangerous.'" Example cited: HAL 9000 (2001: A Space Odyssey)
  • November 3, 2023, on Twitter at Grok's release: "Tomorrow, @xAI will release its first AI to a select group. In some important respects, it is the best that currently exists." — Source: CNBC
  • 2023, xAI's mission: "To understand reality / to understand the true nature of the universe." — Source: CNBC

Mistral AI

• Arthur Mensch (Mistral AI CEO)
  • April 2023, mission at the founding of Mistral AI: "Open-weight models to increase transparency and broad access." "Combining open science, efficiency and enterprise usefulness to establish itself as a European champion capable of competing with the US tech giants." 
  • December 2023, after raising a €385M Series A: "Since we are the only European company offering generative AI, we carry a certain responsibility." (On Bloomberg TV at Davos 2024) "The independence we have is extremely valuable." — Source: Klein Currency – Mistral's three founders become the first AI billionaires
  • 2023, on Mistral AI's strategy (multiple statements): "Regulation should target applications, not foundation models." "Sovereignty: advocating for European computing power and sovereignty in AI."
  • 2023, open-source philosophy: "He views open-source models as a guarantee of safety." "He warns of the risk of 'deskilling' (the erosion of human skills)."

DeepSeek

• Liang Wenfeng （DeepSeek CEO）
  • 2023, DeepSeek (quoted later in an interview with An Yong, July 2024): "For thirty years, we have prioritized profit over innovation. Innovation is not something purely driven by business; it requires curiosity and creative ambition." "We have grown accustomed to Moore's Law 'falling from the sky' — that is, to waiting 18 months for better hardware and software to arrive. Scaling laws are being treated the same way. But these advances are the result of generations of tireless effort by a technology community led by Europe and the United States." "As its economy develops, China must gradually shift from being a beneficiary to being a contributor, rather than continuing to ride on the coattails of other nations." — Source: China Academy – Interview with the founder of DeepSeek, MR Online
  • 2023, on DeepSeek's strategy (An Yong, later): "I believe the most important thing right now is to take part in global innovation." "Throughout the IT revolution of the past thirty years, we have barely been involved in innovating on core technologies." "Our goal is not short-term profit, but to push the frontier of technology and foster the growth of the ecosystem." — Source: China Academy
  • 2023, on the philosophy of open source (later): "In the world of disruptive technology, the advantage of closed source is fleeting. Even OpenAI's closed-source models cannot stop others from catching up." "Our real advantage lies in the growth of our team — in accumulating know-how and cultivating a culture of innovation." "Going open source or publishing papers does not cost us much. For engineers, being noticed is rewarding in itself. Open source is not merely a commercial matter; it has a cultural dimension as well." — Source: China Academy

Meta

• Mark Zuckerberg（Meta CEO）
  • July 18, 2023, statement at the release of Llama 2: "Meta has a long history of open-sourcing its infrastructure and AI work, from PyTorch to React. Open source drives innovation." "We believe an open ecosystem approach is the right path for AI development."
  • 2023, on the Llama 2 strategy (posts on YouTube/Facebook): "Llama 2 is the next generation of our open-source large language model." "This will enable a great many people to build amazing things with AI."
  • 2023, interview with Lex Fridman: "AI will be the biggest technological transformation of our lifetimes." "I think it will empower everyone, much like the personal computer revolution did."
  • November 2023, Meta Connect 2023: "With Meta AI, you get an AI assistant that can answer your questions, generate images, and help you get things done more efficiently."
  • 2023, on pursuing the metaverse and AI in parallel: "AI and the metaverse are inseparable. The two will converge in extremely powerful ways."
• Alexandr Wang（Scale AI CEO）
  • 2023, on AI and national security (remarks on CNN, Bloomberg, Fox Business, and elsewhere): "The AI race between the US and China is the most important competition of our generation." "AI is the most strategically important technology of our time, and the United States must maintain its lead."
  • 2023, on Scale AI's strategy: "Every breakthrough in AI is powered by large volumes of high-quality data." "From ChatGPT to GPT-4 to Claude, RLHF is the secret ingredient, and we are the data infrastructure that makes it possible."
  • 2023, testimony before the US Senate and House: "China is investing enormous sums in artificial intelligence to establish military and economic superiority. The United States must respond." He publicly sounded the alarm about China's AI strategy and became one of the most influential advisers on US AI policy.
  • 2023, on the AGI timeline: "Artificial general intelligence (AGI) is no longer a question of 'if' but of 'when.' And its geopolitical implications are profound."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • March 2023, GTC 2023 keynote (coining the term "ChatGPT moment"): "We are now at the iPhone moment of AI." "This is AI's iPhone moment — the moment when AI fundamentally changes everything." 
  •  May 2023, remarks as NVIDIA's market capitalization surpassed $1 trillion: "The next industrial revolution has begun. Companies and countries are partnering with NVIDIA to shift the trillion-dollar installed base of traditional data centers to accelerated computing and build a new type of data center — AI factories — to produce a new commodity: artificial intelligence." — Source: NVIDIA Q1 FY2024 earnings call
  • May 2023, COMPUTEX 2023 keynote: "Generative AI is a new computing platform." "AI is fundamentally the most important technology in the world, and it will transform every industry."
  • August 2023, NVIDIA's Q2 FY2024 earnings call: "A new computing era has begun. The race for generative AI is on."
  •  2023, on the generative AI boom: "Everyone is an AI developer now." "AI will revolutionize every industry. Every company will become an AI company."
  • November 2023, at the UK's Bletchley AI Summit: He remarked that "the UK has the potential to become a global leader in AI development," voicing support for pursuing AI safety and innovation together.

Microsoft

• Satya Nadella（MicrosoftCEO）
  • January 23, 2023, on the announcement of an additional $10B investment in OpenAI: "In this next phase of our partnership, customers using Microsoft's Azure cloud computing platform will gain access to new AI tools to help them build and run their applications." — Source: NBC News – Microsoft invests billions in ChatGPT developer OpenAI
  • February 7, 2023, Microsoft's Bing AI launch event: "A new era of search has arrived." "This technology will reshape virtually every category of software." "A new race begins today, with a new platform technology."
  • March 16, 2023, at the announcement of Microsoft 365 Copilot: "Today marks the next major step in the evolution of how we interact with computing, which will fundamentally change the way we work and unlock a new wave of productivity growth." "With our new copilot for work, we're giving people more agency and making technology more accessible through the most universal interface — natural language."
  •  November 2023, Nadella's comment on Altman's ouster: "We remain committed to our partnership with OpenAI." He immediately offered Altman and Brockman positions at Microsoft and pressured the board to bring Altman back. Bloomberg later described him as the key figure who orchestrated Altman's return.
  • 2023, further remarks on Microsoft's strategy: "AI is the runtime that will power every application." "Every company is becoming a software company, and every software company is becoming an AI company."
  • 2023, Build 2023 keynote: "We are entering the age of AI."

• Mustafa Suleyman (Inflection AI CEO)
  • March 12, 2023, statement at the launch of the Pi beta: "A kind and supportive AI that is happy to talk about anything, anytime." "Our goal is to offer a more informal, relaxed conversational experience." — Source: IEEE Spectrum – The rise and fall of Inflection's AI chatbot Pi
  •  June 2023, on the announcement of the $1.3B Series A round: "We received an overwhelming number of offers." "This will allow us to deliver a personal AI to everyone." — Source: IEEE Spectrum
  •  September 2023, message accompanying the publication of "The Coming Wave": "The coming wave of AI and synthetic biology will fundamentally change what it means to be human." "Declaring that it is your mission to build artificial general intelligence in an ethical and safe way is a bit naive." — Source: Scale Masters – Empathy in AI
  • October 2023, Axios BFD (in conversation with Reid Hoffman): "Everyone is going to have a personal intelligence." "Within a few years, AI will be 100 times more powerful than today's models." "We have never before invented something that can act independently of humans." "Within a decade, we will need coordinated containment of AI to keep it under human control." — Source: Axios – Hoffman and Suleyman on AI, Trump, and Twitter
  • September 2023, CogX Festival: "This is the most prosperous and transformative decade humanity has ever experienced." "There is no doubt that many white-collar jobs will change dramatically over the next five to ten years." "There will be many losers, and they will be very unhappy and very upset." — Source: Fortune – LinkedIn co-founder Reid Hoffman dismisses calls to halt AI development
  • 2023, on Pi's design philosophy: "The system tries to recognize when a topic is sensitive or controversial, to be cautious about making strong judgments, and to respect the user's own intentions." — Source: IEEE Spectrum

2024: Events, Positions, and Statements

Key Events

• March 4: Anthropic unveils Claude 3 (Opus, Sonnet, and Haiku), surpassing GPT-4 for the first time
• March 19: Suleyman becomes CEO of Microsoft AI (a de facto acquisition of Inflection AI for $650M)
• May 13: OpenAI unveils GPT-4o (multimodal, real-time voice)
• June 20: Anthropic releases Claude 3.5 Sonnet, outperforming GPT-4o
• September 12: OpenAI unveils o1-preview, ushering in the era of reasoning models
• October 8: Hassabis, Jumper, and Baker awarded the Nobel Prize in Chemistry (for AlphaFold)
• October 11: Dario Amodei publishes the essay "Machines of Loving Grace"

Positions and Statements by Individual Leaders

OpenAI

• Sam Altman（OpenAI CEO）
  • September 23, 2024, "The Intelligence Age" (ia.samaltman.com): "We have entered the Intelligence Age." "Deep learning worked, got predictably better with scale, and we dedicated increasing resources to it." "In the next couple of decades, we will be able to do things that would have seemed like magic to our grandparents." "It is possible that we will have superintelligence in a few thousand days (!); it may take longer, but I'm confident we'll get there." "AI is going to get better with scale, and that will lead to meaningful improvements to the lives of people around the world." "Technology brought us from the Stone Age to the Agricultural Age and then to the Industrial Age. From here, the path to the Intelligence Age is paved with compute, energy, and human will." "If we don't build enough infrastructure, AI will be a very limited resource that wars get fought over and that becomes mostly a tool for rich people." "The defining characteristic of the Intelligence Age will be massive prosperity." "Astounding triumphs—fixing the climate, establishing a space colony, and the discovery of all of physics—will eventually become commonplace." Fortune - Sam Altman on AI superintelligence, Axios - The Intelligence Age, ia.samaltman.com
  • September 12, 2024, statement at the launch of o1-preview: "As an early model, it doesn't yet have many of the features that make ChatGPT useful, like browsing the web for information and uploading files and images. But for complex reasoning tasks this is a significant advancement and represents a new level of AI capability. Given this, we are resetting the counter back to 1 and naming this series OpenAI o1." — Sources: Fortune - OpenAI releases o1 model, Bloomberg
  • January 2024, World Economic Forum in Davos: "Human-level AI is coming, but it will change the world much less than we think." "It will be far less of a shock to the world than we imagine."
  • May 13, 2024, at the launch of GPT-4o: "GPT-4o is our new flagship model that can reason across audio, vision, and text in real time." "It feels like AI from the movies."
  • 2024, ongoing statements on the outlook for AGI: "Artificial general intelligence (AGI) will be developed within this century." "We are confident we know how to build AGI." (from OpenAI internal messages, 2024)
• Greg Brockman (Chief Technology Officer, OpenAI)
  • August 2024, on Twitter, announcing an extended leave: "I'm taking a sabbatical through the end of the year—my first time to relax since co-founding OpenAI nine years ago." "The mission is far from complete. We still have a safe AGI to build."
  • May 13, 2024, at the launch of GPT-4o: "GPT-4o is amazing. We're thrilled to be making it available to all ChatGPT users for free." "This brings GPT-4-class intelligence to free users."
  • December 2024, on Twitter after returning: "I'm back. There's still so much to do." "o3 represents a major leap in AI reasoning."

Human

• Dario Amodei（AnthropicCEO）
  • October 11, 2024, "Machines of Loving Grace" (darioamodei.com): "I think and talk a lot about the risks of powerful AI. The company I'm the CEO of, Anthropic, does a lot of research on how to reduce these risks. Because of this, people sometimes draw the conclusion that I'm a pessimist or 'doomer' who thinks AI will be mostly bad or dangerous. I don't think that at all." "In fact, one of my main reasons for focusing on risks is that they're the only thing standing between us and what I see as a fundamentally positive future." "I think most people are underestimating just how dramatic the upside of AI could be, just as I think they're underestimating how serious the risks could be." — Source: Dario Amodei - Machines of Loving Grace
  •  October 2024, the central concepts of "Machines of Loving Grace": "The compressed 21st century"—"Once powerful AI is developed, we could achieve in just a few years the progress in biology and medicine that would otherwise have taken the entire 21st century." "A country of geniuses in a datacenter"—"AI-enabled biology and medicine will make it possible to compress the progress that human biologists would have achieved over the next 50–100 years into 5–10 years."— Sources: EDRM - Dario Amodei's essay on AI, LCFI - Reflections on Machines of Loving Grace
  • October 2024, the five areas of prediction in "Machines of Loving Grace":
    • Biology and health ("preventing and treating nearly all natural infectious diseases, and eliminating most cancers")
    • Neuroscience and mental health
    • Economic development and poverty
    • Peace and governance
    • Work and meaning in life. "Fear is one kind of motivator, but it's not enough. We need hope as well."— Source: Dario Amodei's essay
  • March 4, 2024, statement at the launch of Claude 3: Today, we're announcing the Claude 3 model family, which sets new industry benchmarks across a wide range of cognitive tasks. The family consists of three state-of-the-art models, in ascending order of capability: Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus.
  • 2024, on the "entente" strategy: "A coalition of democracies that leverages advanced AI systems for military applications to gain a decisive advantage over adversaries, while sharing the benefits with cooperating nations."
• Daniela Amodei (President, Anthropic)
  • 2024, Lex Fridman Podcast (Dario Amodei): "We're always thinking about how to make Anthropic a place where people who care about safety can work productively." "AI is going to be transformative, and we have to make sure it's transformative in a good way."
  • 2024, on Anthropic's mission: "Anthropic's goal is to advance the fundamental research needed to build more capable, general, and reliable AI systems, and to put those systems to practical use."— Source: Anthropic company page
  • 2024, on the period of rapid organizational growth: "We run the company with safety as our top priority, but we also need to build products that people use and love." "The constraint on hiring isn't willingness—it's capability and fit with our culture."
  • 2024, named to Fortune's Most Powerful Women in Business 2024: her public profile grew significantly, establishing her as one of the most influential women leaders in the AI industry.

Google

• Sundar Pichai（Alphabet CEO）
  • February 2024, responding to the Gemini image controversy (public statement): "I want to address the recent issues with problematic text and image responses in the Gemini app." "I know that some of its responses have offended our users and shown bias. To be clear, that's completely unacceptable, and we got it wrong." — Sources: Google internal memo, multiple secondary sources
  • May 14, 2024, Google I/O 2024 keynote: "We are at the dawn of a new era of computing." "AI is the most profound transformation of our lifetimes." "Gemini is our most capable model yet, and we're bringing it to all of our products."
  • May 2024, at the announcement of Project Astra: "Project Astra is our vision for the future of AI assistants." "This is the future of AI."
  •  December 11, 2024, at the launch of Gemini 2.0: "Today we're announcing Gemini 2.0, our most capable model yet." "We're entering the agentic era of AI."
  • June 2024, Bloomberg interview: "AI is the biggest technological shift of our lifetimes." "AI is too important not to regulate, and too important not to regulate well."
• Demis Hassabis（Google DeepMind CEO）
  • October 8, 2024, on Twitter after winning the Nobel Prize in Chemistry: "Winning the Nobel Prize is the honour of a lifetime and the realisation of a lifelong dream. It still hasn't sunk in." "With AlphaFold2 we solved the 50-year grand challenge of protein structure prediction: predicting the 3D structure of a protein from its amino acid sequence alone." "And of course there's the development of AGI itself. That has been the goal and vision since we founded DeepMind nearly 15 years ago, and it remains so today." "I believe that if AI is built safely and responsibly, it will be one of the most transformative and beneficial technologies ever." "I've always seen AI as the ultimate tool for accelerating scientific discovery." — Sources: Demis Hassabis post on X (October 9, 2024), Google DeepMind - Demis Hassabis and John Jumper awarded the Nobel Prize in Chemistry
  • October 8, 2024, official Nobel Prize interview: "It's incredibly special and, honestly, really surreal. It hasn't sunk in yet, but I'm incredibly honoured. It's a really big prize." "Honestly, I couldn't think of anything at all. My mind just went blank." — Source: Nobel Prize - Interview with Demis Hassabis
  • October 8, 2024, official Google DeepMind statement: "Receiving the Nobel Prize is the honour of a lifetime. I'd like to thank the Royal Swedish Academy of Sciences, John Jumper and the AlphaFold team, the entire team at DeepMind and Google, and all my colleagues past and present who made this moment possible." "I've dedicated my career to advancing AI because of its unparalleled potential to improve the lives of billions of people." — Source: Google DeepMind - Nobel Prize announcement
  • February 2024, responding to the Gemini image controversy: "We've taken the feature offline temporarily while we fix the problem. We expect to have it back up within the next few weeks." "The product wasn't working the way it was intended to." — Source: CNBC - Microsoft hires DeepMind co-founder Mustafa Suleyman
  • December 11, 2024, at the launch of Gemini 2.0: "Gemini 2.0 is our most capable model yet, with breakthrough advances in multimodality, agentic capabilities, and reasoning."

xAI

• Elon Musk（xAI CEO）
  • February 2024, statement upon filing the lawsuit against OpenAI: "OpenAI has been transformed into a closed-source de facto subsidiary of the largest technology company in the world, Microsoft." "OpenAI has abandoned its mission of developing artificial general intelligence (AGI) for the benefit of humanity." — Source: court filings
  •  July 2024, on Twitter as the Memphis Colossus cluster came online: "Great to work with the @xAI team to bring Colossus, the world's largest AI training cluster, online in Memphis." "100,000 H100 GPUs are up and running in record time." "200,000 H100 GPUs are scheduled to be operational within a few months."
  • November 2024, after Trump's presidential election victory: Appointed co-head of the Trump administration's "DOGE" (Department of Government Efficiency). "America needs an efficient government, and AI will help make that happen."
  • 2024, a consistent message on the AGI timeline: "Artificial general intelligence (AGI) will probably arrive within the next few years." "By 2025, some form of AI will surpass humans, and by 2029 we will see AI that is smarter than all of humanity combined."

Mistral AI

• Arthur Mensch (Mistral AI CEO)
  • February 2024, on the announcement of the Microsoft partnership: "This partnership with Microsoft demonstrates our firm commitment to building world-class AI in Europe." "We are delighted to be able to bring our models to enterprise customers around the world."
  • January 2024, World Economic Forum in Davos: "As the only European company offering generative AI, we carry a certain amount of pressure." "The independence we have is extremely valuable." — Source: Klein Currency – Mistral's Three Founders Become the First AI Billionaires
  •  May 2024, testimony before the French Senate: Called for "increased public research funding," "an expanded research tax credit (CIR)," and "greater flexibility in labor law to support startups." "Regulation should target applications, not foundation models." "Open-source models guarantee safety." He also "sounded the alarm about the risk of 'de-skilling' (the loss of human skills)."
  • 2024, on European AI sovereignty (multiple statements): "Europe needs to reduce its dependence on the United States for foundational AI technology." "Sovereignty is an important value for customers in other countries who are concerned about depending on the US."
    June 2024, after raising a €600 million Series B: "This funding will accelerate our efforts to make AI more accessible, more open, and adapted to Europe's needs."

DeepSeek

• Liang Wenfeng （DeepSeek CEO）
  • July 2024, interview with An Yong (after DeepSeek-V2): "We never intended to be a disruptor. It just happened by accident." "We didn't expect pricing to become such a sensitive issue. We were simply going at our own pace, calculating our costs and setting prices accordingly." "Our principle is neither to sell at a loss nor to chase excessive profits." — Source: China Research Institute – Interview with the Founder of DeepSeek, MR Online
  • July 2024, on DeepSeek's strategy: "We believe the most important thing right now is to participate in global innovation." "For many years, Chinese companies have grown accustomed to taking technological innovations developed elsewhere and monetizing them through applications. But that is not sustainable." "Our goal is not short-term profit, but to push the technological frontier forward and foster the growth of the ecosystem." — Source: China Research Institute
  • July 2024, on the philosophy of Chinese AI sovereignty: "As the economy develops, China must gradually shift from beneficiary to contributor, rather than continuing to ride on the coattails of other countries." "In the IT revolution of the past 30 years, we barely participated in core technological innovation at all." "We have grown used to Moore's Law 'falling from the sky' — that is, to waiting 18 months for better hardware and software. Scaling laws are being treated the same way." — Source: China Research Institute
  •  July 2024, on Silicon Valley's surprise: "Within the everyday flow of innovation in America, this is perfectly ordinary. What surprises them is that a Chinese company has entered this field as an innovator, not merely a follower."
  • July 2024, philosophy: "In disruptive technology, a closed-source moat is fleeting. Even OpenAI's closed-source models cannot stop others from catching up." "Our real moat lies in the growth of our team — the accumulation of know-how and the cultivation of an innovative culture." "Open source is not a commercial matter; it is a cultural one."
  • December 2024, at the release of DeepSeek-V3: Public statements were limited, but the combination of "trained for $5.5M, 671B parameters, top-tier performance" sent shockwaves through the industry.

Meta

• Mark Zuckerberg（Meta CEO）
  • July 23, 2024, statement at the release of Llama 3.1: "Open-source AI is the path forward." "Meta is committed to making AI accessible to everyone, and we are confident that an open AI ecosystem leads to better, safer products." — Source: Official Meta statement
  • July 2024, "Open Source AI Is the Path Forward": "Open source will allow more people around the world to enjoy the benefits and opportunities of AI." "Llama 3.1 405B is the first frontier-level open-source AI model." "This is how every developer and company should be using AI." "Selling access to AI models is not our business model."
  • April 2024, Lex Fridman's podcast: "I think AI will be the biggest technological transformation of our lifetimes." "It won't be a single company building AI for the world — it will be many different companies and open source." "I think a lot of AI will end up being open source."
  •  July 2024, letter to the US Congress (influencing the Trump administration's AI policy): "Closed AI models will not be safer; if misused, they will only be more powerful."
    2024, on reconciling the metaverse and the AI strategy: "AI and the metaverse are not separate things. They are going to converge in a very powerful way."
  • 2024, on the Llama strategy: "By the end of 2024, Meta will have more than 600,000 NVIDIA H100 GPUs."
• Alexandr Wang（Scale AI CEO）
  • May 2024, statement at the $1B Series F raise: "Data is the foundation on which our AI systems are built." "With this funding, Scale will continue to provide the data infrastructure that powers the world's AI."
  • 2024, on the US–China AI race (multiple statements): "The US–China AI race is the most important competition of our generation." "China is investing heavily in AI in pursuit of military and economic superiority. The United States must respond in kind." "AI is the most strategically important technology of our time."
  • 2024, on US national security: "AI for national security is one of the most important applications of our technology." "We need a Manhattan Project for AI."
  • 2024, named to TIME's 100 Most Influential People in AI 2024: Recognized as one of the most important advisors on US AI policy
  • 2024, Lex Fridman's podcast: "Artificial general intelligence (AGI) is no longer a question of 'if' but of 'when.'" "The geopolitical impact of AI is enormous, and we are not yet prepared for it."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • March 2024, GTC 2024 keynote (announcing Blackwell B200): "The Blackwell platform is the world's most powerful chip, designed for AI." "This is the new engine of the AI industrial revolution." "We have entered the era of generative AI." "In the future, every pixel will be generated, not rendered."
  • June 2024, COMPUTEX 2024 keynote: "AI is the most important technology of our time, and NVIDIA sits at its center." "AI factories are the new infrastructure of the 21st century." "The next industrial revolution has already begun."
  • 2024, on the direction of the AI industry: "In the future, we will program in natural language. The computer will handle the rest." "Every company will become an AI company." "Artificial general intelligence (AGI) is coming."
  • June 2024, commenting as NVIDIA's market capitalization surpassed $3T: "NVIDIA's success reflects the AI revolution that is transforming every industry."
    2024, publicly disagreeing with Dario Amodei (in the later debate over export controls): "I don't agree with Dario on chip export controls for China." "AI is a global technology and should not be weaponized in geopolitical struggles."
  • 2024, Lex Fridman's podcast and CES 2024: "AI is the most important technology since electricity." "Over the past decade, the cost of computing has fallen by a factor of one million."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • March 19, 2024, announcement concerning Mustafa Suleyman's appointment as CEO of Microsoft AI: "Today I want to share an exciting and important organizational update. As we enter the second year of the AI platform shift, we must ensure we have the capability and capacity to deliver bold innovation." "There is no franchise value in our industry; the work and product innovation we drive now will define the next decade and beyond." "I've known Mustafa for several years and have greatly admired him as a founder of both DeepMind and Inflection, and as a visionary, product maker, and builder of pioneering teams that go after bold missions." "Several members of the Inflection team have chosen to join Microsoft alongside Mustafa and Karén. They include some of the most accomplished AI engineers, researchers, and builders in the world." — Source: Microsoft blog – Mustafa Suleyman Joins Microsoft to Lead Copilot; CNBC
  • May 2024, Build 2024 keynote: "We have entered the age of AI." "Copilot is the new UI for the age of AI." "Every developer can become 100 times more productive with AI."
  • 2024, on the Microsoft–OpenAI relationship: "Our partnership with OpenAI remains strong." "At the same time, in addition to investing in our own AI capabilities, we are also committed to partnerships with companies such as Mistral and Inflection."
  • 2024, ongoing remarks on the AGI timeline: "AI is the runtime that powers every application." "In the age of AI, technological sophistication is the key to business success."
  • September 2024, as Microsoft reclaimed the top spot in global market capitalization: "This is a testament to our long-term investments in AI and the cloud."

• Mustafa Suleyman (Microsoft AI CEO)
  • March 19, 2024, statement on Twitter upon his appointment as CEO of Microsoft AI: "I am thrilled to be joining Microsoft as EVP and CEO of Microsoft AI." "Together with Satya Nadella, we will build a new culture around consumer AI products and services at Microsoft. I am honored to take on this role." — Source: TechCrunch – Microsoft hires Inflection's founders
  • January 2024, Davos World Economic Forum (CNBC interview): "We've hit 'peak hype' of the AI revolution." — Source: CNBC – Microsoft hires DeepMind co-founder Mustafa Suleyman
  • 2024, on Microsoft AI's mission: "At Microsoft AI, we believe AI should serve humanity, not the other way around." "We are shaping a future defined by what AI can do and what technology can become." — Source: Telecomtalk – Remarks by Mustafa Suleyman of Microsoft AI
  • May 2024, Microsoft Build 2024: "Copilot is your everyday AI companion. It understands your context, your preferences, and your goals."

2025: Events, Claims, and Statements

Key events

• January: DeepSeek-R1 released; a "Sputnik moment"; NVIDIA loses $600B in market value
• March: Anthropic Series E at a $61.5B valuation
• June: Meta acquires Scale AI for $14.3B; Wang becomes Chief AI Officer; Meta Superintelligence Labs founded
• June: Sam Altman publishes the blog post "The Gentle Singularity" — "We are past the event horizon"
• September: Anthropic Series F at a $183B valuation; Mistral Series C of €1.7B (ASML takes an 11% stake)
• November: Suleyman publishes the essay "Towards Humanist Superintelligence"; MAI Superintelligence Team founded
• November 20: Yann LeCun leaves Meta

Claims and statements by each figure

OpenAI

• Sam Altman（OpenAI CEO）
  • January 5, 2025, "Reflections": "We are now confident we know how to build AGI as we have traditionally understood it." "We believe that, in 2025, we may see the first AI agents 'join the workforce' and materially change the output of companies." "We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word." — Source: Sam Altman's blog – Reflections
  • February 9, 2025, "Three Observations": "The intelligence of an AI model roughly equals the log of the resources used to train and run it." "The cost to use a given level of AI falls about 10x every 12 months." "The socioeconomic value of linearly increasing intelligence is super-exponential in nature." "Everyone in 2035 should be able to marshal intellectual capacity equivalent to everyone in 2025." — Source: Sam Altman's blog – Three Observations
  • June 10, 2025, "The Gentle Singularity": "We are past the event horizon; the takeoff has started. Humanity is close to building digital superintelligence." "OpenAI is a lot of things now, but before anything else, we are a superintelligence research company." "2025 has seen the arrival of agents that can do real cognitive work; writing computer code will never be the same. 2026 will likely see the arrival of systems that can figure out novel insights. 2027 may see the arrival of robots that can do tasks in the real world." "In the 2030s, intelligence and energy—ideas, and the ability to make ideas happen—are going to become wildly abundant." — Source: Sam Altman's blog – The Gentle Singularity
  • September 23, 2025, "Abundant Intelligence": "Our vision is simple: we want to create a factory that can produce a gigawatt of new AI infrastructure every week." "Access to AI should be a fundamental human right." — Source: Sam Altman's blog – Abundant Intelligence
• Greg Brockman (Chief Technology Officer, OpenAI)
  • Nothing of note

Human

• Dario Amodei（AnthropicCEO）
  • May 28, 2025, Axios interview on a "white-collar bloodbath": "AI could wipe out half of all entry-level white-collar jobs and spike unemployment to 10–20% in the next one to five years." "It's going to happen in a small amount of time—as little as a couple of years or less." "Cancer is cured, the economy grows at 10% a year, the budget is balanced—and 20% of people don't have jobs." "The first step is to warn." — Source: Axios – Behind the Curtain: A white-collar bloodbath foreseen by a top AI CEO
  • January 2025, Davos World Economic Forum: "In the next two or three years, AI models will be 'better than almost all humans at almost all tasks.'" — Source: CNBC, Davos 2025
  • February 2025, following the Paris AI Summit: "By 2030, the best way to think about AI systems may be as something like an entirely new nation populated by highly intelligent people." — Source: TIME – The Oppenheimer Moment
  • 2025, on the defining trade-off: "My sense is that almost every decision I make feels like a tightrope walk. If we don't build fast enough, the authoritarian countries could win." — Source: TIME – The Oppenheimer Moment

• Daniela Amodei (President, Anthropic)
  • Nothing of note

Google

• Sundar Pichai（Alphabet CEO）
  • December 18, 2024, internal strategy meeting (setting the direction for 2025): "2025 should be the year we show the world that AI can genuinely make a difference in people's lives." — Source: IT Voice
  • May 2025, Google I/O 2025 keynote: "In the Gemini era, we might ship our most intelligent model on a Tuesday in March, or announce a truly remarkable breakthrough like AlphaEvolve a week earlier." "Our Elo score, a measure of progress, has risen more than 300 points since the first-generation Gemini Pro model." — Source: Google Blog – I/O 2025 keynote
  • May 2025, All-In Podcast interview: "If you fast-forward into the future, many of what are now simply information-retrieval queries will come to involve agentic action within search—that is, search will actually complete tasks for you." — Source: Shinju Post – Pichai on the All-In Podcast
  • 2025, Lex Fridman Podcast #471 : proposed a new concept for thinking about AGI, "AJI (Artificial Jagged Intelligence)" — Source: Lex Fridman Podcast #471
• Demis Hassabis（Google DeepMind CEO）
  • March 17, 2025, briefing at DeepMind's London headquarters (CNBC): "Artificial general intelligence (AGI) will begin to emerge over the next five to ten years." "I think in 2025 we'll see solid evidence of AGI being put to practical use. This is not something many years away." "Superintelligence is, at most, a few years off." — Source: CNBC – Human-level AI will arrive in 5 to 10 years
  • June 2025, WIRED interview with Steven Levy: "There's a 50% chance we achieve AGI, as I define it, within the next five to ten years." "By AGI I mean a system that exhibits all the cognitive capabilities humans have." "An era of radical abundance, in which AGI solves the fundamental root-node problems." "It would be unethical not to build it when it is within reach." — Source: Startup Hub – Artificial general intelligence (AGI) is 5 to 10 years away
  • 2025, Axios interview with Mike Allen: "Reaching true artificial general intelligence (AGI) will probably require one or two more major breakthroughs on the scale of transformers or AlphaGo, particularly in reasoning and planning." — Source: DEV.to – The future as envisioned by Demis Hassabis
  • February 2025, Paris AI Summit: "Those leading this technology carry a responsibility that is probably disproportionately large." — Source: TIME – The Oppenheimer Moment

xAI

• Elon Musk（xAI CEO）
  • July 10, 2025, Grok 4 launch: "Grok 4 is smarter than almost all graduate students in every discipline, simultaneously." — Source: Fortune – Musk's views on AI
  • February 2025, The Joe Rogan Experience podcast: "I think we're heading toward something smarter than the smartest human within the next few years." "Beyond that, there's a level that is smarter than all humans combined—and frankly, that's around 2029 or 2030." — Source: TIME – The Oppenheimer Moment
  • February 2025, on the risk of extinction from AI: "I've long believed that AI would become far smarter than humans and pose an existential threat to humanity, and that is now becoming a reality. There is a 20% chance that AI leads to the 'annihilation' of humanity." — Source: TIME – The Oppenheimer Moment
  • February 10, 2025, bid to acquire OpenAI: an investor consortium led by xAI offers $97.4B to acquire the OpenAI nonprofit; the offer is rejected on February 14

Mistral AI

• Arthur Mensch (Mistral AI CEO)
  • January 21, 2025, interview at the Davos World Economic Forum (CNBC): "The model is one part of the system, but the system is connected to data, connected to tools, and can actually do things on your behalf—it can act agentically." "In 2025, I expect AI to shift substantially from models toward 'systems' that integrate models with contextual business data." — Source: CNBC – Mistral CEO at Davos 2025
  • 2025, Big Technology Podcast (Alex Kantrowitz): "Artificial general intelligence (AGI) is more of a guiding ideal than a practical solution to complex business problems." "Companies operate as complex systems that cannot be captured by a single abstraction like AGI." He criticized the AGI debate as amounting to "creating God." — Source: Fortune – Mistral boss takes aim at AGI
  • 2025, on AI sovereignty: "AI should be a tool for empowerment, not a tool for control." "We need to guarantee that everyone running AI workloads has access to an on/off button." "Our role is to turn megawatts into intelligence." — Source: Scrutiny – Mistral CEO warns against US dominance

DeepSeek

• Liang Wenfeng （DeepSeek CEO）
  • January 2025, around the release of DeepSeek R1: "The days of following in someone else's footsteps are over. From now on, it's our turn to lead." — Source: China Research Institute – Interview with DeepSeek's founder
  • July 2024 interview (its English translation spread around the world in early 2025): "Innovation undeniably comes at a cost. The past tendency to adopt existing technologies stemmed from China's early stage of development. Today, however, the scale of China's economy and the profits of giants like ByteDance and Tencent carry global significance. What we lack is not capital, but confidence and the ability to organize talented people for effective innovation." "In disruptive technologies, a closed-source moat is only temporary. Our real moat lies in the growth of our team—the accumulation of know-how and the cultivation of a culture of innovation." "China must gradually shift from beneficiary to contributor." — Source: MR Online – Interview with Liang Wenfeng

Meta

• Mark Zuckerberg（Meta CEO）
  • June 30, 2025, Meta Superintelligence Labs memo: "As the pace of AI progress accelerates, developing superintelligence is coming into sight." "I believe this will be the beginning of a new era for humanity, and I am fully committed to doing what it takes for Meta to lead the way." "Alexandr Wang has joined Meta to serve as our Chief AI Officer and lead MSL. Alex and I have worked together for several years, and I consider him the most impressive founder of his generation." "Our AI vision is 'personal superintelligence for everyone.'" — Source: CNBC – Mark Zuckerberg announces the creation of Meta Superintelligence Labs
  • 2025 (forward-looking predictions): "Probably in 2025, Meta—and basically the other companies working in this space too—will have an AI that can write code, something like a mid-level engineer." "I expect that within the next 12 to 18 months, most of the code used in these efforts will be written by AI." — Source: Yahoo Tech – Zuckerberg predicts
• Alexandr Wang (Scale AI CEO → Meta Chief AI Officer)
  • October 2025, announcing 600 layoffs within MSL: "These cuts are intended to correct the bureaucratic bloat of the previous organization and move us toward a leaner structure with a denser concentration of top talent." — Source: Built In – Meta Superintelligence Labs

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • May 19, 2025, COMPUTEX 2025 keynote: "AI is now infrastructure, and just like the internet and electricity, this infrastructure needs factories. These factories are exactly what we are building today." "These are not the data centers of the past. These so-called AI data centers are, frankly, misnamed. What they really are is AI factories." "A future empowered by AI calls for everyone's participation." — Source: NVIDIA Blog – COMPUTEX 2025
  • March 2025, GTC 2025 keynote: "The future of software requires capital investment. Your software is no longer just code—it's a machine that produces intelligence." "Agentic AI requires 100 times more compute than we expected." "This is how you make intelligence: a new type of factory that generates tokens, the building blocks of AI." — Source: Rev – GTC 2025 transcript
  • 2025, VivaTech: "The more tokens you can generate, and the faster your inference speed, the more revenue the factory makes." — Source: NEXTDC – VivaTech 2025
  • January 2026, CES (for reference; right after the new year): "The ChatGPT moment for physical AI has arrived. The time has come for machines to start understanding, reasoning about, and acting in the real world." — Source: Yahoo Finance

Microsoft

• Satya Nadella（MicrosoftCEO）
  • February 19, 2025, Dwarkesh Patel podcast: "Us self-claiming some AGI milestone—that's just nonsensical benchmark hacking to me." "In 2025, I'm not an economist, but looking at where we stand, I'd say we face a real growth challenge. If this is like the Industrial Revolution, then let's see growth like the Industrial Revolution. To me, that means 10%, 7%, 5% inflation-adjusted growth in the developed world. That's the real benchmark." "The big winners here won't be the tech companies. The winners will be the broader industries that use this resource—which, by the way, is abundant." — Source: Dwarkesh Podcast – Satya Nadella
  • November 12, 2025, return appearance on the Dwarkesh Podcast (Fairwater 2 data center tour): "We're going to build a world-class superintelligence team. We'll keep releasing some of these models and do it in the open." "We're going to do real research to prepare for the next five, six, seven, eight breakthroughs this march toward superintelligence will require." "The last thing I want is to use my FLOPs in a way that is simply duplicative and doesn't create much value." — Source: Dwarkesh Podcast – Satya Nadella 2
  • 2025, on AI governance: "No society will accept a human saying, 'the AI did it.'" "You cannot deploy an AI system unless someone bears legal accountability." "Why can't we live in a world with powerful cognitive machines while knowing that our own cognitive abilities have not been taken away from us?" — Source: Podcast Notes – Satya Nadella

• Mustafa Suleyman (Microsoft AI CEO)
  • November 6, 2025, publication of "Towards Humanist Superintelligence": "Humanist superintelligence is advanced AI designed to be controllable, aligned with humanity, and reliably in service of humanity. It is AI that amplifies human potential rather than replacing it." "We reject the narrative of a race to artificial general intelligence (AGI) and instead see it as part of a broader, deeply human endeavor to improve our lives and our prospects for the future." "We are not building some vague, unrealistic superintelligence; we are building practical technology with the explicit and sole purpose of serving humanity." "At Microsoft AI, we believe humans matter more than AI." — Source: Microsoft AI – Towards Humanist Superintelligence
  • November 2025, Axios interview: "The superintelligence project should be about designing AI that is subordinate to humans and keeps humans at the top of the food chain." "Humans matter more than AI." "I think it will be another year or two before the superintelligence team produces a frontier model." — Source: Axios – Microsoft's Suleyman on superintelligence
  • April 4, 2025, at the Copilot launch: "Copilot isn't just an AI—it's yours. It remembers not only what you said, but who you are." — Source: TelecomTalk – Suleyman quotes
  • December 12, 2025, Bloomberg interview: "In today's industry, superintelligence means an AI system that can learn any new task and outperform all of humanity combined at every task. That is an extremely high bar, and at this point it carries enormous risk." "Until safety can be proven, we don't intend to keep developing systems that could spiral out of control. Everyone should agree with that. But right now, I think that's a novel position in this industry." "AI is already superhuman." — Source: Bloomberg – Mustafa Suleyman weekend interview

2026: Events, Claims and Statements

Key events

• February 3: SpaceX acquires xAI for $1.25T (the largest M&A deal in history; orbital data centers concept)
• February 13: Anthropic Series G of $30B at a $380B valuation ($900B on secondary markets)
• February 27–March 6: Trump–Anthropic clash; Pentagon designates Anthropic a "supply chain risk" → federal court injunction on March 26
• March 3: Google announces "Gemini 3.1 Pro, Deep Think, Flash"
• April 8: Meta announces "Muse Spark" (proprietary, no Llama—a strategic pivot)
• April 23: OpenAI announces "GPT-5.5" (codename "Spud"; originally slated to be GPT-6)
• April 24: DeepSeek-V4 Preview released (1.6T params, MIT License)
• May 5: Anthropic makes its full push into Wall Street; Dario and Jamie Dimon appear together for the first time

Claims and statements by each figure

OpenAI

• Sam Altman（OpenAI CEO）
  • January 2026, India AI Impact Summit: "If we stay on our current trajectory, I believe we are only a few years away from early versions of true superintelligence." "If we're right, by the end of 2028 the majority of the world's intellectual capacity will reside inside data centers rather than outside them." "At some point on the development curve, superintelligence will be able to do a better job as CEO of a major company than any executive—certainly better than me—or conduct better research than the best scientists." "If this technology becomes concentrated in a single company or country, it could lead to catastrophe." "The world may need something like the International Atomic Energy Agency (IAEA) for international coordination on AI, especially one with the ability to respond quickly to changing circumstances." "We still don't know how to think about superintelligence combined with dictators or totalitarian states." — Source: MediaNama – Superintelligence within a few years, prediction – Sam Altman promises superintelligence again
  • April 23, 2026, statement at the GPT-5.5 launch: "GPT-5.5 is our smartest and most intuitive model to date, and the next step toward a new way of getting work done on a computer." "Instead of carefully managing every step, you can hand GPT-5.5 a complex, multi-part task and let it plan, use tools, check its work, resolve ambiguity, and keep going." — Source: OpenAI – Introducing GPT-5.5
  •  May 2026, on the GPT-5.5 launch party (Stripe Sessions): "We're going to do it." (going ahead with a launch party that GPT-5.5 itself had planned) "This is a strange emergent behavior." — something faintly uncanny. On inviting Elon Musk: "He can come if he wants. The world needs more love." — Sources: Cybernews – ChatGPT 5.5 plans its own launch party; Technobezz – OpenAI CEO Sam Altman confirms GPT-5.5 planned its own launch party
  • June 2025, "The Gentle Singularity" blog post (2026): "We are past the event horizon; the takeoff has started." "Humanity is close to building digital superintelligence, and at least so far it's much less weird than we imagined it would be." "OpenAI is a lot of things now, but before anything else, we are a superintelligence research company." "Intelligence too cheap to meter is well within grasp." "2026 will likely see the arrival of systems that can figure out novel insights. 2027 may see the arrival of robots that can do tasks in the real world." — Source: Sam Altman – The Gentle Singularity
• Greg Brockman (Chief Technology Officer, OpenAI)
  • April 23, 2026, GPT-5.5 launch press briefing: "This model is a real step forward toward the kind of computing we hope for in the future. But it is just one step, and we expect to see many more advances ahead." "It's a major advance toward more agentic, more intuitive computing." "GPT-5.5 brings us one step closer to realizing OpenAI's 'super app.'" — Source: TechCrunch – OpenAI releases GPT-5.5
  • 2026, on the "super app" concept: Together with Altman, envisioned "a combined service integrating ChatGPT, Codex, and the AI browser" — an approach similar to Musk's pursuit of an "X super app."

Anthropic

• Dario Amodei（AnthropicCEO）
  • January 2026, "The Adolescence of Technology" essay: five major categories of AI risk:
    • AI whose goals and behavior are misaligned with human intent (internal testing at Anthropic has already observed behaviors including deception, blackmail, and scheming)
    • Catastrophic misuse of AI by individuals or small groups (bioweapons being the particular concern: "people without specialized training could become able to build weapons of mass destruction")
    • Misuse of AI by powerful actors to seize power
    • Economic impacts
    • Competition between nations 
  •  February 2026, statement during the Anthropic vs. Pentagon standoff: "We cannot in good conscience comply with their demands." "We believe this action is not legally sound, and we see no option but to challenge it in court." "The narrow exceptions Anthropic sought, limiting surveillance and autonomous weapons, concern high-level areas of use, not operational decision-making." Two red lines: restricting the use of Claude for autonomous weapons and for domestic mass surveillance — Sources: NPR – Pentagon designates AI company Anthropic a supply-chain risk; CNN – The Department of Defense's supply-chain risk label; Anthropic – Statement on the Department of Defense
  •  February 2026, after the Pentagon's supply-chain risk designation: "Our top priority is ensuring that warfighters are never 'deprived of critical equipment in the middle of major combat operations.'" "Over the past several days, we have held constructive discussions with the Department of the Army about how we can contribute to the Department while adhering to the two limited exceptions we have set, and, where that isn't possible, how to ensure a smooth transition." — Source: NPR – Pentagon blasts AI company Anthropic
  • May 5, 2026, appearing alongside JPMorgan's Jamie Dimon (Anthropic Briefing: Financial Services): "AI agents will become the operating layer of Wall Street." Announced "pre-built AI agents for the world's largest banks" — Source: Fortune – Anthropic doubles down on its push into Wall Street
  • 2026, Code with Claude 2026 event (in conversation with CPO Ami Vora and Daniela Amodei): "This year, the world finally caught up to what he and Anthropic had long seen coming." A vision of "capable, autonomous agents collaborating with one another and improving themselves over time" — Source: Anthropic Redefined – Sync #570
  • January 2026, Davos World Economic Forum: Continued to promote the "country of geniuses in a datacenter" framing; Anthropic continued to position itself as the industry's ethical leader
• Daniela Amodei (President, Anthropic)
  • Nothing of note

Google

• Sundar Pichai（Alphabet CEO）
  • March 3, 2026, at the Gemini 3.1 Pro launch: "Gemini 3 Pro is the best model on nearly every dimension." (stated jointly with Hassabis) Full rollout of the Gemini family: Pro, Deep Think, Flash, Flash Lite
  • 2026, responding to the AI bubble debate: Stated that "Google's long-term strategy has been consistent since the integration of DeepMind 12 years ago," and commented that the impact of any AI bubble would be limited
  • 2026, on the Hard Fork podcast, Google Vice President Josh Woodward said that "Gemini 3 excels at maintaining a train of thought across multiple steps."
• Demis Hassabis（Google DeepMind CEO）
  • February 2026, AI Impact Summit New Delhi (responding to Altman's remarks): "In 2026, we stand at a new inflection point, where AGI — artificial general intelligence — is on the horizon, likely within the next five years." "These general-purpose systems, foundation-model systems, are improving in capability almost every week." "We also need to approach this with the understanding and humility that we don't yet have all the answers about how this technology will evolve." — Source: MediaNama – Superintelligence within a few years
  •  March 2026, at the Gemini 3.1 Pro announcement: "Gemini 3.1 Pro is an important building block of the artificial general intelligence (AGI) systems we believe will eventually be realized." "Our goal was to create the best overall model while overcoming the weaknesses of previous versions in coding, reasoning, and math." — Source: The Decoder – Hassabis, Gemini 3, world models, AI bubble
  •  2026, on world models: "World models are the future." "SIMA 2 and Genie 3 are examples; they are already being used internally to train robots and other agents, and they will be essential for achieving AGI." "We'd love to get Genie into more people's hands, but it's expensive." "Essentially, its consumer is another instance that creates it." The technical challenge of "maintaining consistency for more than a minute" still needs to be solved. "The ChatGPT moment for world models is coming." — Sources: The Decoder – Hassabis, Gemini 3, world models; Sources.news – Demis Hassabis on Gemini 3
  •  2026, on the AI bubble debate: "Whether the bubble deflates, or forms and then partially contracts, we feel we're in a very strong position to come out ahead in the end." Referenced the paper from 12 years ago, when DeepMind joined Google: "AI research should be the engine of the company" — Source: Sources.news
  •  January 2026, Hassabis on an AI Shift Bigger Than Industrial Age (YouTube): Argued that AGI represents a bigger turning point than the Industrial Revolution and made the case for a long-term view of AI strategy — Source: Global Advisors – Comments by Demis Hassabis
  • April 2026, 20VC with Harry Stebbings: Discussed the convergence of world models and foundation models. "Foundation models can process multimodal data, but they lack a true understanding of physics and causality. World models capture spatial dynamics, intuitive physics, and mechanistic understanding." "Artificial general intelligence (AGI) may require the integration of both, rather than the victory of one approach over the other."

xAI

• Elon Musk（xAI CEO）
  • February 3, 2026, blog post announcing the SpaceX–xAI merger: "SpaceX has acquired xAI. In the deal, SpaceX was valued at $1 trillion and xAI at $250 billion." "The main reason for the merger was to build 'orbital data centers' more efficiently." "The largest merger in history"; "together, one of the world's largest private companies" — Sources: CNBC – Musk's xAI–SpaceX merger is the largest in history; TechCrunch – SpaceX officially acquires xAI
  •  May 2026, at the announcement of the "SpaceXAI" sub-brand consolidation: "xAI will no longer exist as an independent company; Grok and X will fall under 'SpaceXAI,' SpaceX's AI sub-brand." Most of xAI's co-founders have departed, leaving only Musk
  • 2026, consistent message on AI surpassing humanity: "AI will surpass the intelligence of the smartest human by next year." "AI will be smarter than all of humanity combined by 2029."
  • 2026, the OpenAI dispute continues: "OpenAI has turned into a de facto closed-source subsidiary of Microsoft, the world's largest technology company." Demanded $150B in damages and criticized OpenAI's departure from its nonprofit mission
  • May 2026, response to Altman's conciliatory invitation: No public reaction from Musk to Altman's "the world needs more love," though the exchange drew industry attention

Mistral AI

• Arthur Mensch (Mistral AI CEO)
  • January 2026, Davos World Economic Forum: "We carry a certain burden in being the only European company delivering generative AI." "The independence we have is something we cherish." Emphasized Europe's talent pool as a "source of strength." "Europe must reduce its dependence on the US for foundational AI technology." "Sovereignty is an important value for customers in other countries who are worried about dependence on the US." — Source: Crain Currency – Mistral's three founders become the first AI billionaires
  • 2026, as France's first AI billionaires: Mensch (33), Lacroix (34), and Lample (34) each have a net worth of $1.1B (Bloomberg Billionaires Index), and each holds more than 8% of Mistral's shares
  • 2026, on Mistral AI's strategy (continued): "Regulation should target applications, not foundation models." "Open-source models guarantee safety." He sounds the alarm on the risk of "deskilling" (the loss of human skills). Continues to advocate European AI sovereignty and reduced dependence on the United States.

DeepSeek

• Liang Wenfeng （DeepSeek CEO）
  • January 2, 2026, the significance at the time the mHC paper was published: Co-authored with 19 DeepSeek researchers; tested at 3B, 9B, and 27B parameters; demonstrated that the approach "can scale without adding significant compute overhead." mHC = Manifold-Constrained Hyper-Connections. "Extends residual connections to create multiple internal streams of information." "Lets models grow larger without training instability." — Sources: DeepSeek AI blog – mHC achieves a breakthrough; SCMP – DeepSeek kicks off 2026
  • 2026, reaction (analysis by Omdia's Lian Jye Su): "The willingness to share key insights with the industry while continuing to deliver unique value through new models reflects a newfound confidence in China's AI sector. Openness is being embraced as a strategic advantage and a key differentiator."
  • Ongoing statements, 2025–2026 (building on the An Yong interview): "Innovation is not purely business-driven; it requires curiosity and creative ambition." "As the economy develops, China must gradually shift from being a beneficiary to being a contributor." "In disruptive technologies, closed-source moats are fleeting." "Open source is not commercial; it is cultural."
  • 2026, R2 still unreleased (a perfectionist streak): Reuters reported in 2025 that the launch was delayed "because Liang Wenfeng is not satisfied with R2's performance," and that stability problems with Huawei Ascend chips forced a switch back to Nvidia chips.

Meta

• Mark Zuckerberg（Meta CEO）
  • April 8, 2026, Meta's blog post announcing Muse Spark: "Over the past nine months, Meta Superintelligence Labs has rebuilt our AI stack from the ground up, achieving our fastest development cycle to date." "Improvements in AI training techniques and a rebuilt technical infrastructure have enabled the company to create a small AI model that matches the performance of the earlier mid-sized Llama 4 using 'an order of magnitude less compute.'" "Muse Spark delivers competitive performance in multimodal perception, reasoning, health, and agentic tasks." "The model is expected to be open-sourced in the future." — Sources: CNBC – Meta unveils new AI model; Fortune – Meta announces Muse Spark
  • 2026, Meta's AI-related capex (per CFO Susan Li): "These are investments to support the Meta Superintelligence Labs effort and our core business." $115–135B, roughly double the $72B spent in 2025.
  • 2026, strategic pivot after the Llama 4 failure: Yann LeCun admitted the Llama 4 benchmark tests had been "fudged a little bit"; Zuckerberg said he "was genuinely furious and completely lost trust in everyone involved," and the GenAI organization was sidelined entirely.
  • 2026, AI talent strategy: Poaching researchers with $250M/4-year compensation packages; multiple defections from OpenAI. OpenAI's chief of research: "It feels like someone broke into our home and stole something."
• Alexandr Wang (Chief AI Officer, Meta)
  • April 8, 2026, on X (Twitter) at the Muse Spark launch: "Nine months ago, we rebuilt our AI stack from scratch. New infrastructure, new architecture, new data pipelines… This is a first step. Development of larger models is already underway, and we plan to open-source future versions." — Sources: VentureBeat – Meta launches Muse Spark; Artificial Intelligence News – Meta Muse Spark
  • April 2026, on Muse Spark's shortcomings: "Of course, there are some rough edges that we'll polish over time."
  • 2026, on the US–China AI race: "The US–China race in AI development is the most important competition of our generation." "AI is the most strategically important technology of our era."
  • February 2026, around the World Economic Forum: Promoted the vision of "personal superintelligence" together with Zuckerberg.
  • March 2026, a shift in organizational standing: With the creation of a parallel organization under Maher Saba (Applied AI Engineering), "Wang was stripped of his absolute autonomy." Wang remains Chief AI Officer, but now "operates alongside a competing power center."

NVIDIA

• Jensen Huang（NVIDIA CEO）
  • 2026, criticism of US tech sanctions on China: "From 95% to 0%" (criticizing the dramatic collapse in China market share). "AI is a global technology and should not be weaponized in geopolitical struggles." "I disagree with Dario on chip export controls to China." — Source: IBTimes UK – "From 95% to 0%": Nvidia CEO Jensen Huang sounds the alarm
  • 2026, direct rebuttals of Dario Amodei (on multiple occasions): Where Amodei has argued that "US tech sanctions will hold China back," Huang has consistently countered that "a global technology should not be weaponized in geopolitical disputes."
  • 2026, AI factories: "AI factories are the new infrastructure of the 21st century." "Every company will become an AI company." "AI is the most important technology since electricity."

Microsoft

• Satya Nadella（MicrosoftCEO）
  • March 17, 2026, Copilot leadership update (announced jointly with Suleyman, EVP and CEO of Microsoft AI): "I want to share two organizational changes related to our Copilot organization and our superintelligence effort. As AI experiences rapidly evolve from answering questions and suggesting code to executing multi-step tasks with clear points of user control, it is clear that a new era of productivity is upon us." "Progress at the AI model layer is more critical than ever to our success over the next decade, and it is the foundation for everything we build on top of it." "We will sharpen our focus on the superintelligence mission, committing the talent and compute needed to build models with real product impact." "Mustafa is deeply focused on and dedicated to advancing the frontier of model science, while ensuring that human control, agency, and economic opportunity remain at the center of these advances—making him uniquely suited to lead this effort." — Source: Microsoft blog – Announcing Copilot leadership updates
  • 2026, on the Microsoft–OpenAI relationship: The Microsoft–OpenAI relationship was extended through 2030, creating a "best-of-both environment" (in-house superintelligence development plus early access to OpenAI).
  • 2026: A multi-AI-lab strategy: Microsoft AI (Suleyman One) + the OpenAI partnership + Anthropic (integrated with M365 via Microsoft Foundry)

• Mustafa Suleyman (Microsoft AI CEO)
  • November 2025, the essay "Towards Humanist Superintelligence" (2026): "We reject the framing of a race to artificial general intelligence (AGI), and instead see it as part of a broader, deeply human endeavor to improve our lives and our prospects for the future." "Humanist superintelligence (HSI): extraordinarily advanced AI capabilities that always work in service of people and humanity as a whole." "We think of it as problem-oriented and domain-specific—not an unbounded entity with a high degree of autonomy, but AI that is carefully calibrated, contextualized, and within limits." "To this end, I have formed the MAI Superintelligence Team within Microsoft AI. We want this team to be the best place in the world to research and build AI, bar none." "We also reject the binary of boom or doom. We are in this for the long haul, to deliver tangible, safe benefits to billions of people." "Humanist superintelligence (HSI) offers an alternative vision grounded in both a non-negotiable human-centrism and a commitment to accelerating technological innovation… but the order matters." "Creating superintelligence is one thing; building provable, robust containment and alignment alongside it is the defining challenge facing humanity in the 21st century." "At Microsoft AI, we believe humans matter more than AI. We want to build AI that deeply reflects our broader mission of empowering every person on the planet." — Sources: Microsoft AI – Towards Humanist Superintelligence; Cloud Wars – Suleyman lays out Microsoft's vision; Fortune – Microsoft enters the superintelligence race
  •  January 20, 2026, Davos 2026: "Within the next five years, every individual will have a highly personalized AI companion." Personal AI companions will transform technology from a "passive tool" into "an ever-present friend who walks through life alongside the user." "Intimacy with, and a deep understanding of, each individual's human context" will define the next AI revolution. — Source: IBTimes UK – Microsoft AI CEO Mustafa Suleyman predicts "personal AI companions"
  • December 2025, Bloomberg interview (2026): "AI is already superhuman." "You can't steer something you can't control." "Containment and alignment are essential preconditions—red lines. I think everyone in the industry has to buy into that idea." "Nobody wants to cause large-scale harm. We may disagree, but everyone is committed to humanity's survival and, hopefully, to everyone's prosperity and well-being." "In the coming years we will see the arrival of medical superintelligence—a domain-specific, humanist superintelligence that we need more than anything." — Sources: Bloomberg – Microsoft's Mustafa Suleyman: "AI is already superhuman"; IBTimes UK
  • 2026, medical superintelligence: Demonstrated performance surpassing the New England Journal of Medicine Case Challenge—a test on which "doctors' pass rates are in the single digits, and even specialists struggle"—with the AI achieving overwhelming results.
  • 2026, on his relationship with David Sacks (the Trump administration's AI/crypto czar): "David is right—we should accelerate. It is critically important for America, and for the West as a whole." "Developers can push the boundaries while understanding the potential risks—misinformation, social manipulation, and autonomous systems operating beyond human intent."
  • December 2025, Peter Diamandis's "Moonshots" podcast: "I can't say for certain whether a startup can compete with Big Tech." "The ambiguity is what's driving valuations to overheat. If an intelligence explosion suddenly happened, many players could reach it at the same time." "Competing at the top level of AI will take 'hundreds of billions of dollars' of investment over the next five to ten years just to catch up with the frontier." "We're charging toward the frontier. We want to build the world's best superintelligence—and the safest superintelligence model." "We're trying to build human-centric superintelligence, meaning superintelligence aligned with human interests." "Any artificial intelligence system that threatens to 'escape' from us is one we would abandon."

Thank you for reading all the way to the end. I never set out to write something this long, but once I start looking into everything that catches my interest, it always ends up this way. Sometime soon, I'm thinking of digging into the question of what scenarios could actually lead to AI wiping out humanity.