Anthropic’s inference margin now funds the model factory at $50M per megawatt

每兆瓦收入与AI模型工厂：Anthropic 如何靠推理毛利反哺训练

Anthropic swung from a −94% gross margin in 2024 to $50M revenue per megawatt in 2026, against a $10–15M compute cost. That inference margin delivered its first profitable quarter: $10.9B revenue and $559M operating profit. Dylan Patel described the loop on the Dwarkesh Podcast—spend $10 on inference, earn $50, then pour the profit into training. The post also cites GLM-5.3-Flash, which matches Claude Opus 4.8 on the Artificial Analysis Intelligence Index with 18B active parameters and a 90–97% cost reduction, showing efficiency boosts profit per megawatt. Nvidia’s $6B Poolside acquisition plus $1B investment bets on turning model building into an industrial process, not artisanal tuning.

Why it matters: Tunguz uses Dylan Patel's data to lay out Anthropic's unit economics clearly: inference margin flipped positive and now funds training, with first profitable quarter in 2026. Solid numbers and fresh angle, but it's secondary analysis, not a primary release — caps below 85.