NVIDIA Plans to Reduce Production of A800 AI GPUs & Focus On H800 For China

By: Muhammad Zuhair
Source: https://wccftech.com/nvidia-plans-to-reduce-production-of-a800-ai-gpus-focus-on-h800-for-china/

NVIDIA is reportedly planning to drastically reduce the supply for its Chinese-oriented A800 AI GPU as it aims at bumping up sales for the higher-end H800.

NVIDIA Wants To Focus Its Production on H800 AI GPUs Instead of A800 For China

The A800 and the H800 GPUs might sound a bit unfamiliar. These GPUs are specifically for the Chinese markets and are cut-down versions of the A100 and H100. They were released to bypass US sanctions in an effort to maintain NVIDIA’s presence in Chinese markets, especially in the artificial intelligence industry.

We previously reported that NVIDIA’s A800 GPUs had witnessed a significant rise in pricing owing to a recent ban rumor. It caused panic in the Chinese markets, which led the GPU to sell at 110,000 yuan/US$15,211, an increase of 10,000 yuan from the launch price. However, it is now being reported that NVIDIA plans to cut the production capacity of A800 to promote sales for its higher-end H800 AI GPU.

SK hynix Lands In NVIDIA's Request To Sample Next-Gen HBM3E DRAM For Future AI GPUs 1

Although the reason behind this move isn’t confirmed yet, it may be that NVIDIA wants the H800 GPU to get more attraction from the Chinese industry to increase sales. The H800 is available for 250,000 yuan and is in the higher price segment; hence it isn’t getting the interest NVIDIA wants. Ultimately, the move would result in higher profits for the company by creating an artificial shortage for the A800, which is already at an all-time high.

The H800 has half the data transfer rate when compared to the H100 and lags behind in training times. It is much behind its chip-to-chip counterpart. However, it is the most high-end offering in the Chinese industry. The H800 is widely popular amongst tech giants such as Alibaba Group Holding, Baidu Inc, and Tencent Holdings since they rival each other to create a ChatGPT-like competitor.

Written by Muhammad Zuhair