US Export Ban Bypassed as RTX 5090 GPUs Reportedly Available in China
High-performance graphics cards reach Chinese market despite restrictions
Reports indicate that NVIDIA's GeForce RTX 5090 graphics processing units (GPUs) are readily available to enterprises in China despite being officially banned under US export policies. According to information shared on Bilibili, a Chinese video platform, these high-performance GPUs are being converted into blower-style cards for deployment in servers to handle artificial intelligence workloads. The source claims that availability of these cards is stable and they can be found on major e-commerce platforms in China.
The RTX 5090 is particularly valuable to Chinese enterprises because it offers superior AI performance compared to the RTX 5090 D, which was specifically designed with reduced capabilities to comply with US export restrictions. The report suggests that even though NVIDIA is preparing to release a new RTX 5090 D V2 model next week, enterprises may prefer the unrestricted RTX 5090 due to its enhanced AI capabilities and larger video memory capacity of 32 GB compared to 24 GB on the compliant model.
KEY POINTS
- •RTX 5090 GPUs available in China
- •US export restrictions bypassed
- •GPUs modified for server use
The modification process involves installing the GPU chip and memory modules on a new printed circuit board (PCB) with a redesigned layout. This conversion transforms the typically bulky custom designs into dual-slot GPUs with blower-style coolers that are more suitable for server deployment. A key modification is the repositioning of the 16-pin power connector from the top to the right side of the card, which reportedly eliminates clearance issues in server chassis and reduces the risk of connector melting problems.
According to the report, an enterprise that has deployed nearly a thousand modified RTX 5090 and RTX 4090 cards claims that not a single unit has experienced connector melting issues. This aligns with previous statements from industry experts who have criticized the top-mounted connector design on high-end GPUs, suggesting that the right-edge positioning is both more space-efficient and safer due to its distance from heat-generating components like voltage regulator modules (VRMs).