Verdict on the original claims: mostly true, with two important corrections and several nuances.
This report verifies each claim in the original narrative against primary and secondary sources, flags errors, and adds documentary context. Sources include court testimony from Musk v. Altman (2026), Walter Isaacson's Elon Musk biography, Sebastian Mallaby's The Infinity Machine: Demis Hassabis, DeepMind and the Quest for Superintelligence, Cade Metz's Genius Makers, and direct interviews given by the principals.
The story is on the record from Hassabis himself, recounted to The Sunday Times and The New York Times.
- Met Thiel at a party at Thiel's California mansion in August 2010, after the Singularity Summit.
- Had "literally one minute" with Thiel; instead of pitching DeepMind, Hassabis asked why chess is such a balanced game and explained the tension between the bishop (3 points) and knight (3 points) having very different powers.
- Thiel was intrigued enough to take a second meeting.
- Thiel's Founders Fund invested ~$1.85M (£1.4M) — one of his first investments outside the US.
- When Google acquired DeepMind in January 2014, Founders Fund owned more shares than the three DeepMind co-founders combined.
Sources: CNBC (Dec 2020), Sunday Times interview, NYT, MoneyWeek profile.
Correction: The Atari game in the demo was Breakout, not Pong. The DQN agent famously discovered the "tunnel" strategy — digging a channel up one side so the ball would ricochet behind the bricks. This was DeepMind's 2013 breakthrough paper.
- Musk and PayPal co-founder Luke Nosek were traveling with Larry Page.
- The clip of DQN learning Breakout caught Page's attention.
- Page reportedly decided then that Google had to own this lab.
Sources: Multiple accounts of the flight; the DQN/Breakout paper is the well-documented underlying technology. The "AlphaGo demo on the jet" framing that appears in some retellings is anachronistic — AlphaGo came later (2016).
This is documented in Sebastian Mallaby's 2026 book The Infinity Machine.
- Acquisition closed January 2014. Reported price: $400–650M (never officially confirmed; most-cited figure ~$500–600M).
- Facebook offered more money than Google. Zuckerberg invited Hassabis to dinner at his Palo Alto home.
- Hassabis applied an "implicit test": he steered the conversation to AI (Zuck enthusiastic), then to VR, AR, and 3D printing (Zuck equally enthusiastic about all of them).
- Hassabis concluded Zuck did not see AI as uniquely transformational — "Facebook offered more money, but I wanted somebody who really understood why AI would be bigger than all these other things."
- Zuckerberg has since confirmed the bidding war publicly, telling South Park Commons: "I did want to buy DeepMind, but they went to Google. Demis was good... he totally did a very good job of playing me off of Google to get a good price."
- Hassabis also extracted a condition: an independent ethics board at Google — the governance precedent Hoffman would later cite for OpenAI.
Musk's last-minute effort to block the deal — TRUE. Per Isaacson's biography, Musk and Luke Nosek tried to cobble together financing to buy DeepMind themselves. They held an hour-long Skype call with Hassabis from an upstairs closet at a Los Angeles party. Musk told him: "The future of AI should not be controlled by Larry." The effort failed.
Sources: Mallaby's The Infinity Machine; Isaacson's Elon Musk biography (excerpted in TIME, Sept 2023); Zuckerberg interview at South Park Commons.
This happened at Musk's 2013 birthday party in Napa Valley, per Isaacson.
- Musk argued that without safeguards, AI could replace humans, making the species irrelevant or extinct.
- Page pushed back: why would it matter if machines surpassed humans, even in consciousness? It would simply be the next stage of evolution.
- Musk: human consciousness is a "precious flicker of light in the universe" that should not be extinguished.
- Page called him a "speciesist" for favoring biological humans over digital minds.
- Musk later confirmed the "speciesist" line in an April 2023 Tucker Carlson interview, calling it the catalyst for co-founding OpenAI.
- The friendship effectively ended. Fortune later reported Page "refused to hang out" with Musk after the DeepMind acquisition.
Sources: Isaacson biography; Musk's own retelling on Tucker Carlson and the Lex Fridman Podcast.
- OpenAI was incorporated December 11, 2015.
- Recruiting dinner took place at the Rosewood Hotel in Menlo Park in summer 2015. Present: Sam Altman, Greg Brockman, Ilya Sutskever, and (arriving late) Elon Musk.
- Musk's famous line at the dinner: "We are summoning the demon… probably within the decade."
- Sutskever's defection from Google Brain was the central recruiting prize. Musk later called it one of the hardest recruiting battles of his life — a back-and-forth with Hassabis on one side and Musk on the other. Sutskever agreed, changed his mind, agreed again.
- Other co-founders: Greg Brockman, Wojciech Zaremba, John Schulman, Andrej Karpathy, Trevor Blackwell, Vicki Cheung, Durk Kingma, Pamela Vagata.
- Initial $1B pledge backed by Musk, Altman, Hoffman, Thiel, Jessica Livingston, AWS, Infosys, YC Research.
- Dario Amodei joined in 2016 from Google Brain (he had co-authored the seminal "Concrete Problems in AI Safety" paper there that same year).
Sources: Court filings in Musk v. Altman (2026); Wikipedia OpenAI founders list; multiple Musk interviews; Cade Metz's Genius Makers.
The "take full control or lose my funding" framing is Altman's version, supported by 2024 OpenAI blog post and 2026 court testimony.
- In 2017–2018, OpenAI was running out of money and the founders debated converting to for-profit to raise more.
- Per OpenAI: "Elon wanted majority equity, initial board control, and to be CEO. In the middle of these discussions, he withheld funding."
- Musk also proposed merging OpenAI into Tesla. He wrote: "Tesla is the only path that could even hope to hold a candle to Google."
- The other founders refused. Musk left the board in February 2018.
- Musk's promised funding largely failed to materialize.
- Reid Hoffman bridged the funding gap to cover salaries and operations during this transition — per OpenAI's own account.
- Altman testified in May 2026 that Musk "needed to have total control over it" and that his departure was actually a "morale boost" for some researchers who disliked his hardcore management style.
- Musk's own December 2018 email: "My probability assessment of OpenAI being relevant to DeepMind/Google without a dramatic change in execution and resources is 0%."
Sources: OpenAI's March 2024 blog post with leaked emails; Musk v. Altman trial testimony (Courthouse News, NPR, CNBC, May 2026).
- Reid Hoffman sat on both the OpenAI board and the Microsoft board — the natural broker.
- Hoffman has confirmed on the Eric Ries Show that he personally introduced Altman to Nadella, and that he abstained from votes on both sides to avoid conflict-of-interest.
- Microsoft invested $1B in July 2019, later expanded to over $13B in additional rounds.
- Bill Gates reportedly told Nadella the initial $1B would be "burned."
- Post-2025 restructuring: Microsoft holds ~27% of OpenAI's for-profit entity (~$135–200B value depending on valuation date), with OpenAI committed to ~$250B in incremental Azure spending.
- During the November 2023 board crisis, Hoffman called Nadella personally to calm Microsoft's nerves (per The Information).
Sources: Eric Ries Show interview with Hoffman; multiple Nadella interviews; The Information (Jan 2024).
Anthropic founding (2021):
- Founded by Dario Amodei (CEO) and Daniela Amodei (President) with seven former OpenAI employees total.
- Other co-founders include: Tom Brown (lead author on GPT-3), Sam McCandlish, Jared Kaplan, Jack Clark, Benjamin Mann, Chris Olah.
- Stated reason for leaving: concerns about AI safety direction post-Microsoft partnership. Dario has since clarified (Lex Fridman, 2024) that it wasn't strictly the Microsoft deal — "it is incredibly unproductive to try and argue with someone else's vision." So he took trusted colleagues and started his own.
- Anthropic is now valued at ~$380B (as of late 2025/early 2026) after raising $30B Series G; major investors include Amazon ($8B) and Google ($2B).
xAI (2023):
- Musk founded xAI in 2023. Grok is its flagship product.
- Musk has also been pursuing the 2024 lawsuit against OpenAI (now in trial as of May 2026 in Oakland) claiming the for-profit conversion violated OpenAI's founding charitable purpose. He seeks $150B in damages.
- OpenAI has countered by releasing internal emails showing Musk himself proposed and supported a for-profit structure — just one with himself in charge.
Sources: Wikipedia Anthropic page; Contrary Research; Fortune profiles; Dario's Lex Fridman interview (2024); Courthouse News coverage of the trial.
1. The talent diaspora. Nearly every major frontier lab traces back to either Google Brain or DeepMind via departures: OpenAI (Sutskever from Google Brain, Amodei from Google Brain), Anthropic (Amodei et al. from OpenAI, originally Google Brain), Inflection (Suleyman from DeepMind), xAI (Babuschkin from DeepMind). The people pool is shockingly small.
2. The governance question. Every chapter turns on what guarantees you can actually bind a frontier lab to: DeepMind's independent ethics board (which Musk concluded was "basically bullsh-t" — its only meeting at SpaceX), OpenAI's nonprofit board (which fired and rehired Altman in five days in November 2023), and Anthropic's Long-Term Benefit Trust. The governance structures keep failing or getting renegotiated under commercial pressure.
3. The five-person network. Hassabis, Musk, Page, Altman, Hoffman — plus Thiel as the early-stage capital connector, Sutskever as the technical kingmaker, and Nadella as the corporate backstop — really did shape ~30 years of tech direction at a handful of dinners, parties, and one private jet flight.
- Walter Isaacson, Elon Musk (2023) — Napa birthday party, "speciesist," DeepMind block attempt
- Sebastian Mallaby, The Infinity Machine: Demis Hassabis, DeepMind and the Quest for Superintelligence (2026) — Zuckerberg test dinner
- Cade Metz, Genius Makers (2021) — OpenAI founding, Sutskever recruitment
- OpenAI's March 2024 blog post "OpenAI and Elon Musk" — primary-source emails 2015–2018
- Musk v. Altman trial transcripts (May 2026, Oakland) — Altman and Brockman testimony on Musk's control demands
- Hassabis interviews in The Sunday Times (2020) and The New York Times — the Thiel chess pitch in his own words
- Reid Hoffman on the Eric Ries Show (2024) — his role brokering the Microsoft-OpenAI introduction
- Dario Amodei on Lex Fridman Podcast (Nov 2024) — his own account of leaving OpenAI
Compiled May 26, 2026. All claims verified against multiple sources. The two corrections to the original narrative: (1) the Atari game shown on the jet was Breakout, not Pong; (2) Dario Amodei has personally rejected the "left over the Microsoft deal" framing as oversimplified, though the timing and direction of the disagreement are correct.