China Rewrites the Rules of AI Acquisition: Singapore Is No Longer a Safe Harbor
Beijing approved Meta's $2 billion Manus deal in December, then reversed it in April. The ruling establishes a new principle: jurisdiction over Chinese-built AI follows where it was made, not where a holding company is incorporated. For every Chinese AI startup that relocated to Singapore to attract Western capital, the message is unambiguous.
Read full story →DeepSeek V4 Runs on Huawei Chips: China's AI Stack Comes Together
DeepSeek's new flagship model officially runs on Huawei's Ascend processors, marking the first deep hardware-software collaboration between the two Chinese AI heavyweights. The milestone isn't just technical: it demonstrates that US export controls on Nvidia chips have succeeded mainly in accelerating China's domestic AI infrastructure.
Read full story →From Blacklisted to Back in the Room: The White House Wants Anthropic Again
Trump called Anthropic a "radical left, woke company" and threatened a six-month phase-out of Claude across federal agencies. Now the White House is drafting an executive action to bring Anthropic back. India is separately seeking its own access to the Mythos model. The "save face and bring them back in" reversal reveals how dependent federal AI workflows had become on Claude in the first place.
Read full story →The Only Winning Move
The encryption wars taught us that you cannot embargo an idea: US export controls on cryptography hobbled American companies, accelerated adversary programs, and chilled international research, while the mathematics proliferated anyway. This week's DeepSeek-Huawei stack and the Manus reversal confirm we are running the same experiment with AI. The only winning move, as WOPR concluded, is not to play. We are playing anyway.
Read opinion →