A modified GeForce RTX 5090 carrying 96GB of VRAM—three times the memory of retail cards—has appeared on Alibaba priced at $3,888. Built by Shenzhen Suqiao Intelligent Technology, the card undercuts both standard pricing and expensive workstation hardware, raising questions about firmware and hardware specifications.
The consumer graphics card market has officially entered uncharted territory. A Chinese manufacturing firm has managed to outfit Nvidia’s flagship Blackwell silicon with an extraordinary amount of video memory, offering a massive hardware upgrade aimed squarely at artificial intelligence developers and power users. This follows a broader trend where Chinese factories have produced modified graphics cards including the GeForce RTX 5080 32GB, the GeForce RTX 4090 48GB, and the GeForce RTX 3090 48GB. While some of these projects are fueled by pure curiosity, most of them are a result of the recent AI boom that has stimulated Chinese factories to think outside the box and give existing and even past graphics cards a memory upgrade so consumers can repurpose them for AI workloads.
Shenzhen Suqiao Intelligent Technology Lists the Modified Card on Alibaba
The modified hardware comes from Shenzhen Suqiao Intelligent Technology Co., Ltd., a Chinese OEM/ODM with over a decade of experience making graphics cards, motherboards, and servers. Rather than a random Chinese chop shop, it is an established manufacturer, which lends some credence to the existence of the GeForce RTX 5090 96GB.
The listed price sits at $3,888, which is roughly 35% less than a standard RTX 5090 currently costs in the U.S. Other Alibaba sellers are apparently asking closer to $5,900 for similar cards, so Suqiao’s price sits on the low end even within this gray market. Even at the higher end of that spectrum, the pricing represents a fraction of what NVIDIA charges for its workstation equivalents.
Clamshell Memory Design and the GB202 Die Connection
Achieving 96GB of VRAM on a consumer-class layout requires radical engineering. The modification likely uses a custom PCB allowing the manufacturer to install the memory chips in clamshell mode, packing GDDR7 modules on both sides of the board to double the total memory on the PCB. This is the same approach NVIDIA itself uses on the Pro 6000.
The connection to enterprise hardware goes deeper than memory layout alone. The RTX Pro 6000 Blackwell, which has 96GB of GDDR7 memory, uses the same GB202 silicon as the GeForce RTX 5090, but with more enabled Streaming Multiprocessors (SMs). Therefore, a 96GB configuration is possible on GB202. The vendor could either procure the GB202 silicon individually or extract it from mainstream GeForce RTX 5090 graphics cards and reball it onto the custom PCB, a method Chinese workshops already used to push the RTX 4090 to 48 GB VRAM.
Contradictory Specifications and Questions of Authenticity
Despite the plausibility of a 96GB configuration on the GB202 die, technical details in the Alibaba listing are full of questionable specifications that have sparked skepticism. The posting mentions GDDR6X memory at 14 Gbps, which doesn’t line up with a genuine 96GB Blackwell build.
Even in clamshell mode, with memory pads on both sides of the PCB, there are only 32 of them, and the RTX Pro 6000 Blackwell has 32 GDDR7 chips, each 24Gb (3GB) in capacity. Micron only produced GDDR6X memory chips in 16 Gb (2GB), so the maximum capacity would be 48GB in clamshell mode. Plus, GDDR6X operates between 19 Gbps and 24 Gbps, much higher than the listed 14 Gbps, which corresponds to standard GDDR6. Running modified memory capacities also demands modified firmware and software-level hacks to get these graphics cards running, similar to how the GeForce RTX 4090 48GB was made possible. Rumors on the street suggest that a leaked firmware for the GeForce RTX 5090 has been available for a few months, which could explain why these cards are starting to surface.
Weighing the Cost Against Enterprise Workstation Alternatives
| Hardware Configuration | Memory Capacity | Approximate Price |
|---|---|---|
| Nvidia GeForce RTX 5090 (Standard) | 32GB | $6,000 (Custom) in the U.S. |
| Modified Suqiao RTX 5090 | 96GB | $3,888 to $5,900 |
| Nvidia RTX Pro 6000 Blackwell | 96GB | $16,000 to $17,999 |
For developers running workloads, memory capacity is crucial. Custom GeForce RTX 5090 graphics cards start at $6,000 in the U.S. market. Meanwhile, after the latest price hikes, NVIDIA is selling the RTX Pro 6000 Blackwell starting at a whopping $16,000, although it can go up to $17,999 depending on the chosen variant.
If legitimate, it would give AI developers a much cheaper way to get large VRAM pools without paying workstation prices. However, buyers would be taking on firmware instability and no official support in return.
