It speaks to China’s AI prowess that the first biography of Sam Altman with access to the subject is this Chinese-language text.

The book is divided into three parts.

Part 1 traces Altman’s journey as a Silicon Valley founder to the early years of OpenAI.

What struck me the most was how closely Altman was integrated into the Silicon Valley ecosystem. Altman first made his name by founding a startup called Loopt, where he partnered with a technical co-founder to create a business to share their location with friends.

Most surprisingly to me, while the business was acquired, Loop likely made very little money for its VC backers. Altman made a personal fortune from this, but Loopt was not a runaway start-up success story.

But the co-founding of Loopt opened Altman to opportunities that would be crucial to his later career. He developed a close relationship with Peter Thiel and Paul Graham, two extremely influential figures in Silicon Valley and defining figures of the post-dot-com crash tech boom.

Crucially, Graham handed over the running of Y-combinator, which had funded start-ups including Airbnb, Dropbox and Stripe, to Altman. This gave Altman unprecedented access to the heart of the Silicon Valley start-up eco-system. As CEO of Y-combinator, Altman had to routinely select and guide start-ups, as well as mediate conflicts between founders and VCs: tasks that he excelled at and the reason why Graham chose him as his successor.

It is under this background that Altman was able to turn his big picture ideas for the development of AI into reality. The key skills (as portrayed in the book) are not technical mastery of any particular subject, but careful cultivation of relationships and the diffusion of tension:

  • ⁠With nothing but his reputation and big picture ideas, Altman coaxed Ilya Sutskever, one of Google’s most talented AI researchers to leave jump ship and join a new outfit.

  • ⁠He then persuaded a range of extremely wealthy Silicon Valley figures (most famously Elon Musk but also LinkedIn founder Reid Hoffmann) to invest/donate huge amounts to OpenAI, which was initially branded a non-profit organization with a mission to ensuring the safe development of Artificial General Intelligence.

  • ⁠For the first few years, despite the talent put together by Altman, OpenAI did not produce any notable results.

  • ⁠It was only a few years in that OpenAI researchers read Google Research’s “Attention is All You Need” and realized “scaling”, or training large-language models on huge amounts of data, would be the fastest way to advancing the progress of AI.

  • ⁠Once OpenAI showed the potential of this approach, Altman then had to do a lot of footwork to get the necessary (and very expensive) computing resources to keep the research going.

  • ⁠The story here is just as much about winner-takes-all competition pressures as it is technical/scientific innovation.

  • ⁠In the biography’s telling, Microsoft leadership were concerned that they were getting left behind in the race of AI. Google was taking a wider and wider lead on (what everyone perceive to be) the next big thing in technology., and Microsoft feared being left behind.

  • ⁠It is for this reason that Microsoft was willing to invest/donate to OpenAI, not least to shake Google’s dominance of the field, rebrand itself as an AI leader, and catch up.

Throughout the narrative, there is a fluent and compelling interweaving of the themes of cut-throat competition, clash between strong personalities, a struggle for political power, and utopian/dystopian visions of how technology will impact the future.

One theme that has surfaced in the news and also forms the subject of Part 2 is to what exact extent OpenAI was to be a non-profit project or a for-profit company.

I remember following the news of Altman’s ousting by OpenAI’s board in late 2023: the missing context (which is expertly supplied by Part 2 of the book) is that Sutskever, the key scientist co-founder of OpenAI became concerned that Altman had departed the roots of the OpenAI project and was developing a technology that would harm humanity. But whatever the merits of the philosophical argument, Sutskever (and his camp) was simply no match against Altman with his connections and public image.

Finally, in Part 3 of the book, Zhou returns to Altman’s childhood and latest, non-OpenAI projects. It completes the image of an ambitious man with innate political nous, well able to navigate the social landscape to push his projects forward.

In some sense, the image of Sam Altman is a return to type of ambitious individuals of earlier ages. As Zhou noted, Steve Jobs had a hippie phase and refused Obama’s initial request to meet. This stands in start contrast with Altman, Thiel and Musks’s willingness to leverage the state to shape society. The age of the apolitical “builder” founder may be a passing phrase.