Premium: What If...We're In An AI Bubble? (Part 3)

Ed Zitron's third installment of his AI bubble series lays out the financial math behind why the current AI investment cycle may be unsustainable. Key arguments: NVIDIA needs ~$435B in annual compute demand to justify its GPU sales, but real demand outside OpenAI and Anthropic is minimal. OpenAI and Anthropic have made over $1.1 trillion in compute commitments they can only honor if their revenues grow to $184B and $174B respectively by 2029 — projections that look increasingly unrealistic. OpenAI posted a negative 122% operating margin in Q1 2026 and ChatGPT growth has stalled. Enterprise customers like Uber are already struggling to justify AI token spend with measurable ROI. Anthropic's revenue growth is partly driven by 'token-maxxing' behavior that companies are beginning to pull back on. The piece concludes that both companies face existential financial risk if their growth trajectories slow even modestly, and previews a final installment covering data center debt collapse, private credit write-offs, and potential deaths of OpenAI and Anthropic.