The AI Compute Crunch Is Here—And It's Reshaping the Economy
Key Takeaways
- ▸Major AI companies are implementing rationing: GitHub paused signups and removed expensive models, Anthropic restricted Claude Code, and OpenAI shut down Sora due to insufficient compute
- ▸The compute shortage is cascading through the entire economy: GPU, RAM, and storage prices have tripled; consumer electronics are becoming scarce; and data center power demands are stressing electricity grids
- ▸Venture-backed AI pricing is unsustainable—companies cannot endlessly run products at a loss. The current model mirrors Uber's early strategy of burning investor cash to suppress prices and gain market share
Summary
The age of venture-capital-subsidized, artificially cheap AI is ending as major tech companies grapple with an unsustainable compute crisis. AI companies are implementing aggressive rationing across the board: GitHub paused Copilot signups and removed expensive model access, Anthropic restricted Claude Code availability in lower-tier plans, and OpenAI's CFO has repeatedly warned of insufficient capacity—prompting the company to shut down products like Sora entirely. The pattern is clear: companies can no longer afford to run their AI products at a loss.
The crunch extends far beyond product restrictions into the broader economy. Consumer hardware prices have skyrocketed as manufacturers redirect production toward AI infrastructure—the same external SSD that cost $159 a year ago now costs $575, and essential components like GPUs and high-capacity storage are nearly unavailable. Prices for Microsoft 365, Notion, Salesforce, and Google Workspace have all increased, while electricity costs in AI data center hubs have risen sharply, prompting some states to actively block new data center construction. Even water supplies are becoming a concern.
This trajectory mirrors Uber's venture-capital-subsidized model of the early 2010s, where investors burned billions to artificially suppress prices and dominate markets. That strategy proved unsustainable, eventually forcing dramatic price increases and eroding both rider and driver value. The AI industry now faces the same reckoning: continue burning investor cash to subsidize products, sharply raise prices to reflect true costs, or reduce service quality. None of these options are palatable, but one of them is inevitable.
- Startups relying on affordable AI tokens will see costs rise dramatically, eliminating the arbitrage that made replacing human employees with AI economically attractive
- The shift from cheap AI to premium, rationed AI services is accelerating; consumers and businesses should expect significant price increases and reduced access in the coming months
Editorial Opinion
The compute crunch exposes a fundamental delusion in tech investing: the belief that exponential demand can forever outpace supply when the underlying infrastructure is finite. Like Uber, AI companies bet that investor capital could defy economic gravity indefinitely—but there is no such thing as an infinite supply of chips, electricity, or capital. The reckoning is here, and the damage is spreading: startups built on cheap-API arbitrage will collapse, consumer electronics are being strangled by data center competition, and the era of abundant, cheap AI is ending. Brace for the enshittification phase.
