Fable fight exposes AI access risk

Summary

Deirdre Bosa reports on how the White House export fight with Anthropic exposes a new risk for AI developers: access to closed models is not guaranteed. The spat strengthens the case for open source AI, especially Chinese models like DeepSeek and Alibaba's Qwen, which can be downloaded and customized without cancellation risk. Investors see winners in open model companies, model routers, and neutral AI infrastructure, while closed model labs face a changing landscape ahead of their IPOs.

• Anthropic's Fable shutdown highlights risk that advanced AI models can be pulled or restricted by policy.
• The incident gives open source AI its strongest argument, as models can be owned and customized.
• Chinese open models such as DeepSeek, Alibaba's Qwen, GLM, and MoonShots are gaining adoption.
• A Chinese AI player (GPU) and a Hong Kong AI stock (Z AI) surged overnight on the news.
• Potential winners include open model companies, model routers, and neutral AI infrastructure.
• The irony is that U.S. controls meant to protect AI leadership may push users toward Chinese-controlled systems.
• This mirrors the chip export control saga, which accelerated China's own technology stack development.