Premium: How Much Money Does AI Need?

Source: Where’s Your Ed At Date: 2026-08-14 URL: https://www.wheresyoured.at/premium-how-much-money-does-ai-need/

Summary

Zitron argues the AI industry’s spending trajectory through 2030 is untethered from any plausible revenue path: he cites OpenAI’s reported $750 billion in compute commitments through 2030 against $1.65 trillion in off-balance-sheet obligations and debt held across hyperscalers, and accuses AI boosters of “number-blindness” — accepting figures of that size without asking whether they can ever be repaid.

Implications

The bear-case counterweight to the month’s capability headlines. Every open-weight jump (DeepSeek-V4-Pro’s cyber and coding gains, Qwen’s ecosystem dominance) and every closed-lab feature release assumes the capital keeps flowing; this piece is the reminder that the spending underneath all of it is a bet, not a given. Worth pairing with tracked-dep pricing signals (Gemini 3.7 Flash’s price-halving, DeepSeek’s MIT-licensed give-away) as evidence of margin compression happening in real time against Zitron’s thesis that revenue isn’t catching up to capex.