Anthropic and OpenAI Have Begun The Subprime AI Crisis

Source: Where’s Your Ed At Date: 2025-07-07 URL: https://www.wheresyoured.at/anthropic-and-openai-have-begun-the-subprime-ai-crisis/

Summary

Ed argues Anthropic and OpenAI have initiated a “subprime AI crisis” by operating at massive losses while charging customers artificially low prices — Anthropic projecting $3B in losses for 2025 against $4B in annualized revenue. When these companies are eventually forced to raise prices to approach sustainability, the ripple effects will hit every company that built products on top of their APIs at current pricing. The crisis isn’t about model quality; it’s about a pricing structure that cannot persist and a market that built on it as if it could.

Implications

• AI financial sustainability. The subprime analogy is precise: just as mortgage-backed securities hid underlying credit risk through repackaging, current AI pricing hides the underlying unit economics by accepting massive losses. The repricing event will be sudden and painful for everyone downstream.
• Generative AI ROI. Every business case built on current API pricing needs a stress test at 2x-5x pricing — that’s the realistic range once labs stop subsidizing adoption. Most enterprise AI projects don’t survive that test, which means the current deployment wave is built on a temporary price floor.
• Capital markets. $3B losses on $4B revenue is a -75% operating margin. Venture capital is absorbing that gap on the bet that scale changes the economics. Ed’s claim is that the economics don’t improve with scale — inference costs scale with usage — making the VC bet structurally unsound.