The OpenAI Bubble

Source: Where’s Your Ed At Date: 2026-07-15 URL: https://www.wheresyoured.at/the-openai-bubble/

Summary

Ed Zitron’s essay argues the entire AI industry’s capital expenditure is contingent on OpenAI specifically, and that OpenAI’s finances don’t support the scale of commitment built around it: ~$852B in planned spending through 2030, $748B in remaining performance obligations to Microsoft/Amazon/Oracle, and a plan to spend $50B+ on compute in 2026 alone — against revenue and subscriber-conversion numbers he says can’t cover it (projected 80% Plus-subscriber churn, ad-revenue projections exceeding the entire market’s estimated size). He documents specific hyperscaler exposure — Oracle’s credit outlook downgraded over a $300B build-out contract, SoftBank’s negative outlook tied to its OpenAI stake — and frames several high-profile partnership announcements (Samsung/SK Hynix wafer deal, AMD, Disney/Sora) as having inflated stock prices without materializing as described.

Implications

• AI-bubble/economics thread: this is the most detailed single-source case yet for a circular-financing collapse scenario — hyperscalers funding the labs that are simultaneously their biggest claimed future customers — with named, cross-checkable financial exposures (S&P downgrades, specific contract dollar figures) rather than general skepticism.
• Cascading dependency risk: the argument names concrete secondary victims if OpenAI falters (CoreWeave, Cerebras, Oracle, NVIDIA demand) — worth tracking against real earnings/downgrade news rather than taking the piece’s framing at face value.
• Counter-read needed: this is opinion/advocacy journalism (Zitron has run this thesis for over a year); treat the specific numbers as claims to verify against primary filings, not settled fact, before citing them elsewhere in the radar.