Google put $40B into Anthropic on a Friday.

DeepSeek dropped a new flagship model the same day.

One year after the original DeepSeek shock that wiped a trillion dollars off US AI stocks, they did it again. Quietly. No drama. Just a new model.

The thesis everyone forgot: model performance is converging. Open-weights, low-cost Chinese models are 6 months behind frontier and closing.

If you're a Fortune 500 CIO, the question stops being "which lab wins."

It becomes "why am I paying frontier prices for tasks an open-weight model handles?"

Nobody at Google, OpenAI, or Anthropic wants you asking that question. So they're spending hundreds of billions on compute, distribution, and consulting alliances to make sure you don't.

The model is becoming a commodity. The integration is the only premium.

That's been my thesis for two years. April 24 just made it cheaper to prove.