The high video memory (VRAM) demands of AI models and new workloads are making older graphics cards popular again in the hardware world. AiGPU, a Japan-based hardware repair shop, has started offering a professional modification service that increases the VRAM capacity of NVIDIA’s old flagship GeForce RTX 2080 Ti model from 11 GB to 22 GB.

This process, which costs approximately $282 (45,000 Japanese Yen) if users send their own RTX 2080 Ti graphics cards, doubles the card’s memory capacity. Hardware experts remove the standard 1 GB GDDR6 memory chips on the printed circuit board (PCB) and solder 2 GB modules in their place.

This conversion process, which is not limited to just the physical soldering process, also involves special adjustments to the card’s memory configuration resistors and BIOS so that the graphics processor (GPU) can recognize and address the new memory blocks without any problems. AiGPU offers modification services for existing cards as well as selling ready-made models that are directly upgraded to 22 GB.

The memory upgrade processes offered by the store are not limited to the RTX 2080 Ti only. With the modifications made, the memory limits of other well-known GPU models are also exceeded:

This hardware modification does not increase the raw computing power of the graphics processor or the memory data bus width. Therefore, an RTX 2080 Ti with 22 GB of VRAM does not provide a direct frame rate (FPS) increase compared to the original model in standard gaming scenarios.

However, doubling the memory capacity makes the card an extremely economical workstation alternative in areas experiencing high VRAM bottlenecks, such as large local language models (LLM), visual production tools, and 3D rendering applications. Heavy projects that suffer from errors or slowdowns due to insufficient memory can be processed seamlessly thanks to the expanded memory pool.