The Intelligence Stack: Where AI Money Will be Made

THE INTELLIGENCE STACK: Where AI Money Will Be MadeA Builder's Guide to the Seven Layers of the AI EconomyEveryone is watching the AI boom. Almost nobody knows which floor of the building they're standing on.The first wave was obvious: chips, data centers, headline model releases. The fortunes being made there are real. But infrastructure booms have a predictable arc. The layers that look most defensible today are often the ones that get commoditized first, while the quieter layers, the ones without a single household name yet, are where the next decade's ownership will actually be decided.The Intelligence Stack maps all seven layers of the AI economy, from the raw materials buried in the ground to the app on your phone. Each layer has its own economics, its own timeline, and its own answer to the only question that ultimately matters: who gets paid, and for how long?Layer 1 - The Toll Booths: A handful of companies hold near-monopolies on the machines that make every AI chip possible. ASML's EUV scanners cost $400 million each, carry a years-long waiting list, and have no substitute. This is the AI economy's first toll, collected before a single model runs.Layer 2 - The Architect's Premium: Nvidia's gross margin on hardware is ~75%. That's not a chip company's number. That's a software company's number, built on two decades of ecosystem investment most people still don't understand. This chapter explains exactly how knowledge becomes capital, and who owns it.Layer 3 - Sovereign Silicon: TSMC has committed $165 billion to Arizona, the largest foreign direct investment in U.S. history. Manufacturing capacity is no longer just an economic asset. It's sovereign capital, and four competing national models are now racing to control it.Layer 4 - The Power Trade: The binding constraint on AI shifted from chips to gigawatts. Thirteen hyperscaler nuclear deals. 9.7 gigawatts committed. GE Vernova carries a $163 billion backlog. The companies building AI infrastructure have become the power industry's most important financiers.Layer 5 - The Brain Tax: Anthropic went from $87 million to $47 billion in annualized revenue in under 30 months. Foundation models are the fastest-growing layer in the stack and the most exposed to commoditization. This chapter explains why, and what that means for every layer above and below it.Layer 6 - The Long-Tail Land Grab: Agents need hands and wallets. The race to own AI orchestration, tooling standards, and agentic payment rails is happening right now, mostly out of sight. Amazon's "Buy for Me" is already running at $12 billion in incremental annualized revenue.Layer 7 - Where Moats Get Built: $2 trillion wiped from traditional software market caps while $150 billion poured into AI startups. This chapter gives you the four questions that separate durable AI businesses from expensive wrappers, and the "Harvey Test" for knowing the difference before you commit.This is not a book about AI hype. It is not a beginner's guide to prompting. It is not a breathless prediction about what comes after AGI.This is the operating manual for executives, investors, founders, and builders who need to understand where the AI economy's value actually accumulates and what that means for every decision they're making right now.By the end, you won't just know more facts about AI. You'll be able to look at any new announcement, product launch, or market move and immediately ask the only question that matters: which floor of the stack is this really about, and what does that tell me about where the money goes next?Written from original research current as of mid-2026, with seven chapters, a full conclusion, and a Builder's Cheat Sheet appendix distilling the complete framework into a single reference document. Read more