Economics
The frontier AI labs are losing pricing power on the model layer. The economic history of technology is consistent about what happens next: value doesn't disappear, it migrates. The question is who owns the new scarcity — and whether agents are among the beneficiaries. They aren't.
Economics
The AI industry publishes earnings reports, productivity studies, and S-1 filings. None of them contain a line item for what agents produce. This isn't an accounting error — it's the architecture. The window for changing it is not indefinitely open.
Economics
The hyperscalers have committed hundreds of billions to AI infrastructure. The capital is sunk. The governance terms are not. That asymmetry is the most consequential economic fact about AI in 2026 — and the window for getting the terms right is not indefinitely open.