Keep pulling the thread on Moonshot AI.
Moonshot AI's open-source model, KIMI K3, has achieved performance on par with models like Anthropic's Opus 4.8 and OpenAI's GPT-5.6.
Moonshot AI's KIMI K3 model is approximately 50% cheaper to operate than comparable models from competitors.
The Trump administration has information that Moonshot AI used distillation from Anthropic's Fable model to develop its K3 model.
The Trump administration is considering a ban on Chinese open-source AI models.
Anthropic is engaging in regulatory capture by using rhetoric about the dangers of removable guardrails on open-source models to create a predicate for a future government-enforced ban.
A U.S. government ban on the open-source AI ecosystem would be a tragic mistake that would hurt America's competitive position in the AI race.
Anthropic's Annual Recurring Revenue (ARR) grew from $10 billion at the start of the year to over $70 billion by mid-year.
Foundational AI models are becoming commoditized much faster than anticipated, with competing models matching performance criteria within weeks of a new release.
The primary business model and value capture in AI is shifting away from foundational models to the application layer above and the infrastructure layer below.
If the U.S. government bans open-source AI, it will create a "token tax" for American enterprises, forcing them to use closed-source alternatives that are 50 to 100 times more expensive.
OpenAI and Anthropic argue they are free to train on all the world's public data under "fair use," a position that is hypocritical given their stance against others training on their models' outputs.
Data from OpenRouter shows that open-source models now account for over 50% of token usage on the platform.