Nvidia's GPUs remain the best solutions for AI training, but Huawei's own processors can be used for inference.

Investors are jittery over three big concernsChina's DeepSeek shaking up AI demand, a potential shift from GPUs to ASICs, and ...

Google Cloud is now offering VMs with Nvidia H100s in smaller machine types. The cloud company revealed on January 25 that its A3 High VMs with H100 GPUs would be available in configurations with one, ...

Huawei Chairman Howard Liang announced that 2024 revenue exceeded CNY860 billion (approx. US$118.6 billion) at the Guangdong ...

TL;DR: DeepSeek, a Chinese AI lab, utilizes tens of thousands of NVIDIA H100 AI GPUs, positioning its R1 model as a top competitor against leading AI models like OpenAI's o1 and Meta's Llama.

So, DeepSeek (and other Chinese AI firms) used Nvidia H800 GPUs, which are watered-down H100 GPUs, to meet export regulations. These aren't as powerful, so DeepSeek's engineers had to figure out a ...

which consists of 2,048 H100 AI GPUs. For those that don't know, NVIDIA touts Blackwell's performance as 2.5x the floating-point performance of Hopper, the previous generation of AI GPUs.

Nvidia's GPUs remain the best solutions for AI training, but Huawei's own processors can be used for inference.