A Stanford Lecture Explains Why AI Value Gets Trapped In Chips
NOBODY TALKING: Everyone's racing to build AI apps. Nobody's talking about who actually keeps the money.
Why it matters
The economics of generative AI fundamentally break the software scaling playbook—unit economics degrade with user growth because GPU costs scale linearly, meaning value concentrates in semiconductor supply rather than application layers. This reshapes how founders should think about defensibility and margin structures in AI businesses.
The key facts
4 to knowTraditional software margins improve with scale; generative AI margins degrade with each additional user due to GPU compute costs
Application layer effectively subsidizes semiconductor monopolies
Semiconductor suppliers capture disproportionate share of AI revenue growth relative to app-layer creators
Published Apr 25, 2026 (academic/strategic framework piece, not breaking news of an event)
Go to the source
Forbes Innovationforbes.com
Publisher excerpt: The traditional software playbook promised that more users meant better margins. Generative AI breaks that rule entirely; every new user requires burning expensive GPU compute, which means the application layer is effectively subsidizing a semiconductor monopoly that collects the lion's share of…