DeepSeek is a Chinese AI that claims to be more cost-efficient than US competitors. Its open source chatbot has propelled it to the top of the App Store in over 51 countries. Now it’s been revealed that the app runs on a Huawei chip.
DeepSeek R1 LLM was trained on Nvidia’s H100, but uses the Ascend 910C for inference. Inference is the process of using the trained model as a response generator.
pic.twitter.com/6IAgQlQ3ou
— Alexander Doria (@Dorialexander) January 28, 2025
The information comes from @Dorialexanderwho points out that Ascend chips are not dealing with training, so the GPU power requirements are not that high. pic.twitter.com/6IAgQlQ3ou
— Alexander Doria (@Dorialexander) January 28, 2025
The information comes from@Dorialexanderwho points out that Ascend chips are not dealing with training, so the GPU power requirements are not that high.
The Ascend 910C is not suitable for training due to its lower performance. Huawei will address this issue in the upcoming 920C chipset, which is designed to compete with Blackwell B200 – the leading Nvidia chip for AI operations. The DeepSeek iOS app
