Key takeaways
- Read four books yourself this year: Power and Prediction, Co-Existence, The Skill Code and one vendor history. Co-Existence, The Skill Code and The AGI Chronicles together run to 912 pages, so with Power and Prediction the set takes about an hour a week for a quarter.
- Delegate the technical books, but read their opening chapters. AI Engineering and Vibe Coding give you the vocabulary to question your teams' plans; your senior engineers need the rest.
- Read at least one history of the frontier labs. Your AI roadmap depends on a few companies whose governance and incentives these books explain better than any vendor briefing.
- Bring AI Snake Oil to procurement and The Coming Wave to the board. One sharpens your scepticism of product claims, the other frames the risk conversation directors expect.
- Older strategy books still work. Power and Prediction (2022) and Competing in the Age of AI (2020) describe operating-model changes that agents make more urgent.
Which AI books are worth a technology executive's reading time?
Ranked within each group by how much a technology leader gets out of it. "Checked" says where each number came from and when.
Read yourself: strategy and the board
-
Ajay Agrawal, Joshua Gans and Avi Goldfarb · Harvard Business Review Press · 2022
For: CTOs, CIOs and CAIOs who own the AI portfolio and present it to the board.
This is the book that explains why your pilots are not moving the P&L. The authors separate point solutions from system redesign, and that distinction is the most useful lens available for sorting an AI portfolio before a budget round. Read all of it; it is the one strategy book directors are likely to have heard of.
Checked: Publisher store (HBR Store: 15 Nov 2022), 19 Nov 2026
-
Marco Iansiti and Karim R. Lakhani · Harvard Business Review Press · 2020
For: Technology leaders arguing for a shared data and platform layer rather than scattered AI projects.
Iansiti and Lakhani describe the firm as an AI factory, with data pipelines, models and experimentation at the core. Written before generative AI, it is still the clearest argument for treating AI as operating architecture rather than a feature, which is the case a CTO has to make for platform investment. The case-study chapters can be skimmed.
Checked: Publisher store (HBR Store AI books listing: 7 Jan 2020), 19 Nov 2026
-
Ethan Mollick · Portfolio · 2026 (October) · 240 pages
For: Executives leading AI adoption beyond engineering, and anyone briefing a CEO on how work is changing.
Mollick's new book deals with agents that complete whole tasks and are sometimes better than the employee supervising them. It is the book I would give a CEO or CHRO who asks what AI means for their organisation in practical terms. At 240 pages it fits into a week of commutes.
Checked: Publisher page (Penguin Random House: Portfolio, 20 Oct 2026, 240 pages), 19 Nov 2026
-
Matt Beane · Harper Business · 2024 · 224 pages
For: Engineering leaders deciding how many junior engineers to hire when agents write much of the code.
Beane's research shows that automation often removes the very tasks through which novices become experts. For a CTO, that is the hidden cost of cutting graduate hiring: in five years there is no one ready to become a senior engineer. Read it before the next headcount plan, not after.
Checked: Retailer listing (Porchlight: Harper Business hardcover, 11 Jun 2024, 224 pages) and author site, 19 Nov 2026
Read yourself: the vendors you depend on
-
Kevin Roose · Farrar, Straus and Giroux · 2026 (October) · 448 pages
For: CAIOs and CTOs whose roadmaps depend on OpenAI, Anthropic or Google models.
Roose's account of the three frontier labs, drawn from more than 150 interviews and internal documents, is effectively a briefing on the governance, culture and incentives of your most important suppliers. That context matters when you assess model deprecation risk or negotiate an enterprise agreement. If you read one vendor history, make it this one.
Checked: Retailer listing (Porchlight: Farrar, Straus and Giroux hardcover, 6 Oct 2026, 448 pages); Macmillan page blocked automated access, 19 Nov 2026
-
Sebastian Mallaby · Penguin Press · 2026 (March) · 480 pages
For: Leaders running on Google Cloud or Gemini, and anyone tracking AI for scientific R&D.
Mallaby's biography of Demis Hassabis explains how DeepMind's research priorities sit inside Google's commercial ones. Leaders with Gemini or Vertex in their stack get a clearer view of how that supplier thinks than any account team will give them. Optional if you have already read The AGI Chronicles; useful if Google is your primary AI vendor.
Checked: Publisher page (Penguin Random House: Penguin Press, 31 Mar 2026, 480 pages), 19 Nov 2026
-
Karen Hao · Penguin Press · 2025 (paperback May 2026) · 496 pages
For: CIOs and CAIOs who must answer to boards, regulators or staff about the ethics of their AI suppliers.
Hao's critical history of OpenAI covers governance failures, data labour and the resource costs of training. Those are the questions an ESG committee or a works council will eventually ask about your AI estate, and this book prepares you for them. Pair it with Roose for balance.
Checked: Publisher page (Penguin Random House: hardcover 20 May 2025, paperback 19 May 2026, 496 pages), 19 Nov 2026
-
Stephen Witt · Viking · 2025 · 272 pages
For: CTOs and infrastructure leaders signing GPU, cloud or neocloud commitments.
Witt's history of Nvidia explains why one supplier sets the price and availability of AI compute. That is background every leader negotiating capacity should have. Short enough to finish on a long flight; the paperback is due in February 2027.
Checked: Publisher page (Penguin Random House: Viking hardcover 8 Apr 2025, 272 pages; paperback listed for 23 Feb 2027), 19 Nov 2026
-
Chris Miller · Scribner · 2022 · 464 pages
For: Leaders whose AI plans are exposed to export controls, Taiwan risk or hardware lead times.
Miller's history of the semiconductor industry is the context behind every board question about supply-chain concentration and export rules. It does not cover generative AI, and it does not need to. Skim the early history and read the second half closely.
Checked: Retailer listing (Porchlight: Scribner hardcover, 4 Oct 2022, 464 pages); Simon & Schuster page blocked automated access, 19 Nov 2026
Skim, then hand to your team
-
Chip Huyen · O'Reilly Media · 2024 (December) · 534 pages
For: Read chapter 1 and the evaluation chapters yourself; give the whole book to staff engineers and AI platform leads.
Huyen's book is the reference text for building on foundation models, and the evaluation chapters explain why so many AI features ship without a reliable quality measure. A leader who has read those chapters asks better questions in architecture reviews. Budget for a team copy and a reading group.
Checked: Publisher page (O'Reilly: December 2024, 534 pages), 19 Nov 2026
-
Gene Kim and Steve Yegge · IT Revolution · 2025 (October)
For: VPs of Engineering rolling out coding agents; read the parts on team practice, delegate the rest.
Kim and Yegge write about agent-assisted development as a change in team workflow, with checkpoints, small batches and guardrails. The leadership chapters help when setting policy on how much agent-written code reaches production without human review. The publisher lists a 2026 Axiom Book Awards gold medal.
Checked: Publisher product page (IT Revolution: 21 Oct 2025, Axiom gold medal), 19 Nov 2026
-
Sebastian Raschka · Manning · 2026 (June) · 440 pages
For: Heads of AI and ML platform leads; executives only need the introduction.
Raschka's book shows in code how reasoning models are built, including why they spend more tokens per answer. That cost mechanism is what your FinOps team sees on the invoice. Hand it to whoever owns model selection and inference spend.
Checked: Publisher page (Manning: June 2026, 440 pages), 19 Nov 2026
Keep for risk and governance
-
Arvind Narayanan and Sayash Kapoor · Princeton University Press · 2024 · 360 pages
For: CIOs and procurement leads evaluating AI vendor claims, especially in HR, risk and decisioning.
Narayanan and Kapoor give a practical test for which AI claims deserve belief, and their critique of predictive systems that score people applies directly to HR and risk tooling. I would make it required reading for anyone who signs AI purchase orders. The chapters on predictive AI are the ones to read first.
Checked: Publisher page (Princeton University Press: 24 Sep 2024, 360 pages), 19 Nov 2026
-
Mustafa Suleyman with Michael Bhaskar · Crown · 2023 (paperback 2025) · 352 pages
For: Executives who brief boards and regulators on AI risk.
Suleyman's containment argument is the version of AI risk that directors and policymakers have most often encountered. Knowing it lets you meet the board where it is before steering the conversation to the operational risks you actually manage. Skim the synthetic biology chapters unless you work in life sciences.
Checked: Publisher page (Penguin Random House: Crown hardcover 5 Sep 2023, paperback 1 Apr 2025, 352 pages), 19 Nov 2026
How much AI reading should a technology executive do?
About an hour a week is enough, provided it goes to books rather than more feeds. Newsletters and podcasts tell you what changed this week; books give you the frameworks that decide which of those changes matter. At an hour a week, a leader gets through four or five serious books a year, which is exactly the size of the personal reading list above.
The common mistake is reading at the wrong altitude. A CTO does not need to finish a 500-page engineering text, and a CAIO does not need three histories of OpenAI. Pick one book from each group in this list, read the strategy books in full, and treat the technical titles as chapters to sample and then delegate.
Pair the books with a short weekly source so the reading stays current. Our list of AI newsletters for technology leaders covers that slot, and the university programmes and courses page is the next step for leaders who want structured study.
Which AI books help with board conversations?
Power and Prediction gives you the language for value: why some AI investments change the business and most only change a task. That framing works well in an investment paper, and it pairs with a structured AI business case when you need the numbers behind it.
For risk, directors tend to arrive with ideas from The Coming Wave or from press coverage of books like If Anyone Builds It, Everyone Dies. You do not have to agree with either. Knowing the arguments lets you acknowledge the concern and then move the discussion to the risks you can govern: data leakage, model failure, vendor dependence and regulatory exposure. Our AI governance guide covers that operational agenda.
The vendor histories help in a different way. When a board asks how dependent the company is on one model provider, The AGI Chronicles and Empire of AI give you the governance history to explain why that dependence is a strategic risk, and our framework on AI vendor capture risk turns it into a plan.
Which books should you delegate, and to whom?
Delegate by accountability. AI Engineering goes to staff engineers and the AI platform lead, who own evaluation and architecture. Vibe Coding goes to engineering managers rolling out coding agents, since it is mostly about team practice. Build a Reasoning Model (From Scratch) goes to whoever owns model selection and inference cost, because that is where reasoning models change the bill.
Delegation works best with a deliverable attached. Ask the reading group to return with one change to the evaluation process, the coding-agent policy or the model-selection criteria. That turns a book budget into an operating improvement, and it gives you a reason to read the summary chapters yourself.
Keep AI Snake Oil closer to the centre. Procurement, legal and HR leaders who approve AI purchases benefit from it as much as engineers, and it is the cheapest protection against buying a product whose claims do not survive testing. The guide to buying AI is the practical companion.
Which 2026 AI books matter most for executives?
Three of this year's releases earn a place on an executive's shelf. Co-Existence is the most current account of how agents change knowledge work, The AGI Chronicles is the most complete briefing on the frontier labs, and The Infinity Machine fills in Google's side of the story. All three were published between March and October 2026, so they reflect the market you are actually buying in.
Build a Reasoning Model (From Scratch), published by Manning in June, matters for a narrower reason: it explains the cost profile of reasoning models, which spend far more tokens per answer than standard chat models. I would not read it personally, but I would make sure the person who owns the AI budget has.
What do AI books miss for technology leaders?
Almost none of them deal with the economics of running AI at scale. Token costs, capacity planning, GPU contracts and the trade-off between seats and API usage are absent from the popular books and only touched on in the technical ones. Those decisions need current data, which our pages on compute capacity planning and frontier-model access cost try to provide.
They also say little about organisation design. The Skill Code is the exception on talent, but none of these books tells you where AI engineers should sit or how agents change team topology. For that, I would go to our AI team design material and to Thomas Prommer's essay on the future of software engineering, then use the books to test the plan against first principles.
Listed here, or should be?
If your entry is out of date or you think something belongs on this list, email [email protected]. Entries are editorial and free. A featured, labelled Sponsored slot is available through vendor listings.
Related questions
More from AI Education
- AI Education: where technology leaders learn AI in 2027
- AI conferences
- AI YouTube channels
- AI accounts on X
- AI newsletters
- AI podcasts for executives
- AI voices on LinkedIn for executives
- AI peer communities for executives
- AI courses and university programs
- Agentic engineering resources
- Vibe coding resources
- AI tool stack for CTOs
- The full AI Education hub on We The Flywheel, with niche lists for every role