Premium: The AI Compute Demand Story Is A Lie

2026-05-04 · Source: Ed Zitron's Where's Your Ed At · Field: Finance & Economics — Capital Markets & Investment Management, Economic Analysis & Policy · Depth: Intermediate, medium

Summary

The prevailing narrative of insatiable AI compute demand is challenged as a fabrication driven by hyperscaler subsidies rather than genuine market need. Amazon and Google have pledged a combined \$65 billion more into Anthropic, which previously raised \$30 billion and projects significant losses (\$11 billion in 2026 and 2027) despite ambitious revenue forecasts. Microsoft similarly invested over \$13 billion in OpenAI, later allowing it to diversify compute providers while retaining a 20% revenue share. The author estimates that over 70% of Microsoft's \$37 billion annual AI run rate and 80% of its 2GW AI capacity are consumed by OpenAI's \$24 billion Azure spend. Amazon's \$15 billion annual AI revenue is similarly dominated by Anthropic's \$12 billion AWS spend. Hyperscalers have collectively spent \$803 billion in AI capex, with OpenAI and Anthropic raising over \$252 billion, largely for compute. The article contends that outside these few entities, actual AI compute demand is minimal, below \$1 billion, and the capacity crunch stems from data center construction timelines, not organic demand.

Key takeaway

For investors evaluating AI sector growth, you should critically assess claims of "insatiable demand" and differentiate between organic market pull and hyperscaler-driven subsidies. Your due diligence must scrutinize the true source of AI revenue and compute spend, recognizing that much of the reported activity stems from a few large players funded by their cloud providers. Be wary of circular financing models that inflate market size, as this could indicate systemic weakness in related infrastructure providers.

Key insights

The "AI compute demand" narrative is largely an illusion created by hyperscaler subsidies to a few key AI companies.

Principles

• Hyperscalers heavily subsidize AI startups to create captive compute demand.
• Data center construction timelines, not demand, drive capacity constraints.
• Circular financing schemes inflate perceived AI market activity.

In practice

• Scrutinize AI revenue claims, especially from hyperscaler-backed entities.
• Evaluate AI infrastructure investments against actual, diverse market demand.

Topics

• AI Compute Demand
• Hyperscaler Investments
• Anthropic
• OpenAI
• Data Center Capacity
• AI Market Valuation

Best for: Investor, Consultant, Tech Journalist

Related on AIssential

• Premium: AI's Circular Psychosis — The AI economy's perceived growth is an illusion, sustained by circular funding and compute demand from two unprofitable companies.
• Premium: What If...We're In An AI Bubble? (Part 2) — The AI industry's financial structure relies on unsustainable growth, concentrated demand, and weak due diligence, signaling a potential bubble.
• If enough other companies report the same, the bubble pops. 🫧 — Major companies are reporting poor AI return on investment and reliability issues, suggesting a potential market correction for overvalued AI ventures.
• OpenAI exec says company hopes to burn $50B of somebody else's money on compute this year — OpenAI expects to spend $50 billion on compute in 2026, largely through conditional "investments" from partners.
• AI Weekly Issue #494: SpaceX wants $80 billion. OpenAI wants a trillion. — Public markets are now scrutinizing the high valuations of major AI companies, shifting from private to public pricing.
• Sam Altman’s OpenAI Empire Cracks — $200B Needed to Survive 2030 — OpenAI faces significant financial strain, requiring vast capital while its ecosystem shoulders substantial debt and profitability remains uncertain.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Ed Zitron's Where's Your Ed At.