Loading...

The US Commerce Department has closed the cloud loophole for AI chips, immediately disrupting multimillion-dollar data center deals across Asia. This move forces Chinese tech firms to face a double hit: frozen physical chip shipments and the loss of overseas cloud compute capacity that they previously used to skirt sanctions.
According to The Decoder, the new rules require cloud providers in third countries like Singapore, Malaysia, and Indonesia to obtain a license from the Bureau of Industry and Security (BIS) to serve Chinese customers. Existing operators have a 90-day transition period but must rigorously screen clients for ties to China. The licensing threshold is low, equivalent to roughly 300 modern AI chips, while Chinese labs typically require several thousand processors.
The Information reports that Beijing-based startup Moonshot AI canceled a $150 million contract with a Southeast Asian cloud provider in Indonesia. Moonshot had planned to rent Nvidia Blackwell Ultra B300 chips to train its next model, Kimi K4, and had agreed to prepay 50 percent of the cost. Additionally, Bloomberg reports that five lenders are considering dropping out of a $480 million loan for Singapore-based cloud provider Aolani, which was set to supply chips in Malaysia for ByteDance.
Physical supply chains are also affected. The Financial Times reports that TSMC and Samsung stopped certain processor shipments to Chinese customers while applying for licenses. This impacts not just AI chips but also advanced smartphone processors for manufacturers like Xiaomi, Oppo, and Vivo. A New York Times investigation revealed that Alibaba acted as the secret financier behind Alice Huang's Singapore-based firm Megaspeed, which bought over $4 billion worth of Nvidia chips. Megaspeed operated these chips in a Malaysian data center, routing data to China, with some chips reportedly smuggled physically. This scandal, involving links to military-adjacent research at Zhejiang Lab, prompted the aggressive tightening of controls.
For small and mid-sized companies, this news is a stark warning against relying on single-source, high-risk infrastructure. While your company likely does not deal in billions of dollars of compute, the principle remains: if your AI strategy depends on hardware or cloud services from jurisdictions subject to geopolitical friction, your roadmap is vulnerable. The collapse of the Moonshot deal shows that even pre-paid contracts can be voided overnight by regulatory shifts. For a small team, this means that "cloud-washing" or using third-party proxies to access restricted or unstable infrastructure is a liability, not a solution. You need transparency in your supply chain. If your vendor is a mid-tier provider in a region with complex export compliance issues, you are exposed to the same 90-day transition risks that are currently halting major industry deals. Stability and compliance are now more valuable than raw price cuts on compute.
Google's Gemini desktop app can now control macOS to handle multistep tasks like file organization and form filling, available to US subscribers.

Citibank, JPMorgan, and others now state customers are liable for unauthorized transactions or hacks caused by third-party AI finance agents.

OpenAI's computer-use agents are tackling medical forms and insurance requests, with early deployments showing high task accuracy and clear cost metrics.

Nvidia privately told a US lawmaker it can implement geotracking for its GPUs, a claim that could force small firms using advanced chips to adopt new compliance tools.

Comments1
What about the impact on Chinese firms' ability to access AI chips through domestic cloud providers, are they exempt from these new rules?