A new version of the Nvidia RTX 5080 featuring 32 GB of GDDR7 VRAM has appeared in China.
Although Nvidia’s standard reference model launched with 16 GB, the Asian market—led by Chinese manufacturers and hardware modifiers—has started offering this “boosted” version to meet the growing demand for more power in Artificial Intelligence tasks.
This 32 GB variant uses high-density GDDR7 memory modules (3 GB per chip), allowing it to double the original capacity without changing the base Blackwell GB203 architecture.
According to reports, these cards are being sold mainly in blower-style designs, ideal for workstations and servers where multiple GPUs are installed in parallel. This setup improves heat dissipation by pushing hot air out of the case.
Focus on AI, Not Just Gaming
While 32 GB of VRAM sounds like a dream for gamers, the real target of this model is the AI and deep learning sector.
Because of the shortage of dedicated server chips and trade restrictions, many Chinese companies are turning to these modified RTX 5080 units. With double the memory, the GPU can handle larger language models (LLMs) and complex datasets that the standard 16 GB version cannot process efficiently.
Even though these current versions are modified or built for specific markets, rumors suggest that Nvidia is closely watching this development.
It’s possible that a future RTX 5080 SUPER could officially come with more memory. For now, however, the 32 GB version remains a unique import-market product.
The standard RTX 5080 still ships with 16 GB of GDDR7 memory, but a SUPER model with 24 GB or 32 GB might become a reality soon.
