NVIDIA has removed more than half of its authorized AI chip customers in Asia, targeting smaller cloud providers and distributors in Singapore, Malaysia, and Japan. The move is part of a broader effort to prevent advanced semiconductors, including the Blackwell series, from reaching China through indirect routes. The purge follows enhanced compliance checks and comes after US officials warned that Chinese entities are acquiring restricted processors via intermediaries in Southeast Asia.
Why the customer purge happened
The Commerce Department issued guidance in May specifically aimed at shipments to overseas subsidiaries linked to Chinese firms. Smuggling networks and transshipment schemes have moved significant chip volumes into China, often at premium black-market prices. NVIDIA’s decision sacrifices short-term revenue to protect compliant markets and avoid penalties under US export rules. Affected companies can adjust their internal controls and reapply for approval later, but for now they’re off the list.
Impact on smaller cloud operators
Larger firms with strong compliance frameworks are more likely to keep their spot on the approved list. Smaller cloud operators face immediate supply disruptions. The purge could slow the spread of cutting-edge AI infrastructure beyond the major hyperscalers. That means companies that rely on NVIDIA’s chips to run AI workloads may have to look elsewhere or scale back plans.
DeepSeek’s in-house chip development
DeepSeek is developing its own AI chip focused on inference, aiming to reduce its reliance on both NVIDIA and Huawei. But the project remains early-stage. It’s unclear when or if the chip will reach production. For now, DeepSeek, like many others, depends on NVIDIA’s supply chain — a chain that just got a lot narrower.
The question now is whether smaller firms can tighten their compliance fast enough to get back on NVIDIA’s list, or whether they’ll be forced to find alternative suppliers in a market where options are shrinking.
