$30B in revenue - and it's a warning sign

Reader Outcome: After reading this edition, the reader will understand why commoditization of the model layer is not an accident but the collective strategy of everyone else in the stack — and think differently about where durable returns in AI actually accrue.

• A reusable test for judging whether any AI company’s margins are owned or merely borrowed

• Why the fastest-growing software revenue in history coexists with collapsing pricing power — and why both are true

• The strategic logic (”commoditize your complement”) that explains why hyperscalers, chipmakers, and open-weight labs are all, deliberately, pushing model prices toward zero

• What the fiber-optic bust of 2001 predicts about who banks the value of abundant intelligence

• Where value is actually pooling right now: distribution, workflow context, verification, and power

• How to reposition — as an investor, operator, or builder — for a world where the miracle is cheap and its complements are scarce

1. Executive Summary

2. Deep Dive in One Sentence

3. Why This Topic Matters Now

4. The Big Question

5. The Conventional Narrative

6. What’s Really Happening

7. The Economics Behind the Shift

8. Winners and Losers

9. Second-Order Effects

10. Strategic Implications

11. Mental Model of the Week

12. Key Takeaways

13. Closing Thought

• Frontier labs are posting historic revenue growth — roughly a $30B annualized run rate at the leader by April 2026, a figure Salesforce needed two decades to reach — while the price of the intelligence they sell collapses underneath them.

• This is not competitive misfortune. The model layer is the complement of every other layer in the stack, so the rational strategy of chipmakers, hyperscalers, open-weight labs, and enterprise buyers alike is to make models cheap.

• The labs must participate in their own commoditization: their growth depends on price elasticity, so they sell deflation to expand volume — liquidating the scarcity that justifies their valuations.

• Durable value is migrating to what abundance makes scarce: distribution, workflow context, verification, and energy. Enterprise AI budgets grew from $1.2M to $7M annually between 2024 and 2026, but that spending increasingly flows to inference and application infrastructure rather than model API fees.

• The strategic question for every AI business is no longer “how good is the model?” but “who profits when this layer gets cheaper — and can they make it cheaper?”