Empire of AI Review: Karen Hao’s Critical History of OpenAI and the ChatGPT Era

Books of AI is reader-supported. When you buy through links on our site, we may earn an affiliate commission at no extra cost to you. As an Amazon Associate we earn from qualifying purchases.

Technical books explain how large language models work. Very few explain who builds them, how the money flows, and who carries the costs. Karen Hao’s Empire of AI is an attempt to tell that story, centered on OpenAI and its chief executive, Sam Altman.

Hao has reported on artificial intelligence for years, including at MIT Technology Review, where she profiled OpenAI long before ChatGPT made it famous. The book was published by Penguin Press in May 2025 and has drawn more than 2,400 ratings on Amazon. We looked at what it covers, how readers respond and where it fits on an AI reading list.

What the book covers

The narrative follows OpenAI from its founding as a nonprofit research lab to its transformation into one of the most valuable companies in technology. Along the way it covers the decision to scale up language models aggressively, the commercial partnership with Microsoft, the launch of ChatGPT, and the dramatic boardroom crisis of November 2023, when Altman was fired and then reinstated within days.

Hao does not stop at the company’s offices. A large part of the book follows the supply chain behind AI: the contract workers who label and filter training data, and the communities affected by the data centers that train and run the models. Those chapters are what set Empire of AI apart from the many business histories of the AI boom.

The central argument

The title is the thesis. Hao argues that the leading AI companies operate like empires, claiming data, labor, land, energy and water on a global scale in pursuit of an ever larger model, and justifying it with a vision of benefits to come. She contrasts that “scale at all costs” path with other ways AI could be built.

You do not have to accept the argument to find it useful. It gives readers a framework for questions that come up in any discussion of generative AI: who benefits, who pays, and who decides.

What readers say

Empire of AI averages 4.4 out of 5 from more than 2,400 ratings. Amazon’s AI summary of customer reviews paints a clear but divided picture:

• Content: many call it a must-read, with well-researched insights into the AI industry and OpenAI’s inner workings.
• Writing and storytelling: the writing quality is praised, with one reader noting it makes AI comprehensible to non-experts, and the narrative is described as gripping.
• Readability and balance: mixed. Some find it extremely readable, while others consider it not balanced.
• Bias: several readers point to the author’s political biases.

That split is the main thing to know before buying. The book has an argument to make, and readers who want an even-handed corporate history may be frustrated.

How it compares

Supremacy by Parmy Olson covers overlapping ground with a different structure, following the rivalry between OpenAI and Google DeepMind. It is shorter, a little less polemical, and readers rate it highly as informative and well researched. If you read only one, choose Hao for depth and a critical lens, Olson for a brisker overview.

For a view from inside the industry, Mustafa Suleyman’s The Coming Wave, which we cover in our review of The Coming Wave, argues that AI must be contained, but from the perspective of a founder rather than a critic. Older classics such as Human Compatible by Stuart Russell address the long-term safety questions that OpenAI’s founders often invoked.

Why it belongs in the LLM section

Empire of AI contains almost no technical explanation of transformers or training. It belongs on a list of LLM books anyway, because it explains why these models were built the way they were: why scale became the dominant strategy, why the models are trained on as much data as possible, and why a handful of companies control the frontier. Readers who combine it with a technical book, such as Hands-On Large Language Models, get both halves of the picture.

Who should buy it

Buy Empire of AI if you work in or around technology, marketing, policy or media and want to understand the industry shaping your field, or if you are a student looking for a well-reported case study of a company at the center of a technological shift. Skip it if you want neutral reporting or a technical primer, and budget time: at nearly 500 pages, it is a commitment.

Alternatives worth considering

Frequently asked questions

What is Empire of AI about?

It is a reported history of OpenAI, from its founding as a nonprofit lab to the ChatGPT era, including the November 2023 firing and reinstatement of Sam Altman. Hao argues that leading AI companies behave like empires, drawing on resources, data and labor from around the world.

Is Empire of AI biased?

It takes a clear critical position, and Amazon’s summary of reviews notes that several readers see political bias. Others consider that critical lens the book’s value. Read it alongside a more industry-friendly account if you want balance.

Do I need technical knowledge to read it?

No. It is written for general readers, and buyers specifically mention that it makes AI comprehensible to non-experts.

Who is Karen Hao?

A technology journalist who has covered AI for years, including as a senior AI editor at MIT Technology Review, where she reported on OpenAI well before ChatGPT launched.

How we review: we research every product using the manufacturer’s specifications, independent expert sources and patterns across verified customer reviews, then score it against the alternatives in its price class. Prices and availability change often, so always check the current listing on Amazon.