Empire of AI

Empire of AI is a work of investigative narrative that treats today’s most influential AI companies—especially OpenAI—not as quirky startups or neutral research labs, but as the command centers of a new kind of power. Karen Hao’s central idea is that the modern AI boom resembles empire-building: it is expansive, extractive, and justified by grand stories about progress. The book follows the rise of OpenAI from its idealistic origin story to a sprawling enterprise that depends on enormous capital, vast physical infrastructure, and hidden human work distributed across the globe.

Rather than describing AI as an inevitable technological wave, the book insists that what happened was the result of choices made by particular people under particular incentives. The “race” to artificial general intelligence (AGI) is portrayed not as destiny, but as a strategy—one that encourages risk-taking, secrecy, and concentration of control. Hao shows how the pursuit of “frontier” capability became the organizing principle for everything else: corporate structure, product launches, safety decisions, lobbying priorities, and even the story OpenAI tells about itself.

What emerges is a portrait of an industry that sells a future of abundance while operating through familiar patterns: resource extraction, labor exploitation, and political capture. The book does not argue that AI must be abandoned. It argues that AI as currently built is not the only possible version of AI—and that the reigning model has costs that are systematically shifted onto the least powerful.

Founding Myth: From Mission-First Idealism to Competitive Dominance

OpenAI’s early identity, as presented here, is almost purpose-built to sound like a moral counterweight to Big Tech. Founded as a nonprofit with high-profile backers, it pledged enormous funding and framed its goal as building AGI for the benefit of everyone. The founding language emphasized openness and a willingness to cooperate—even to step aside—if another group was closer to success. A major motivation was the fear that a single company could dominate AGI, particularly a giant with deep resources and a strong head start.

Hao argues that this moral framing mattered because it functioned as legitimacy. It positioned OpenAI as a public-spirited institution rather than another profit machine. But within a few years, the organization increasingly came to resemble the thing it claimed to be preventing: a powerful, secretive entity determined to win.

After internal tensions and the departure of key figures, the project began drifting toward commercialization. Financial reality played a role, but the book also emphasizes status, ego, and competitive urgency. The desire to “get there first” became a moral argument in itself: if OpenAI didn’t win, someone worse would. That logic can justify almost anything—especially opacity and acceleration.

The transformation accelerated with a structural shift: OpenAI reorganized into a “capped-profit” model that allowed it to raise huge capital while still claiming mission alignment. That move unlocked investments on a scale a nonprofit could not support, most notably a billion-dollar deal with Microsoft. But it also changed the DNA of the organization. Transparency became selective. Collaboration became conditional. Research became product-driven. The book treats this not as a small ...