Deepseek, which has recently made a name for itself, has led to major fluctuations in global markets with its new LLM model, and in a short time, Chatgpt was the most low -downloaded artificial intelligence application. This rapid progress of Deepseek has caused some questions about the equipment used by the company during the education process. However, uncertainty began to find answers.

My inference gets its power from Huawei

As mentioned above, Deepseek created a shock effect in the US technology sector. Moreover, it is claimed that it has achieved this with a budget of only two months and less than $ 6 million, and the equipment used became a matter of curiosity. R1’s current answers come from the special server systems where Huawei Ascend 910C chips are used.

For the kidnappers, Huawei’s chip is seen as a direct rival to Nvidia’s Hopper H100 series. However, its relatively low performance limits its suitability for education.

The detailed calculation sources used for Deepseek R1 AI model are not specific for now and there is a lot of misunderstanding. First, the “$ 5 million” figure reflects the cost of running the latest model rather than the total training cost. Secondly, on the agenda that Deepseek has reached more than 50,000 from the H100s of Nvidia.

It is certain that the use of Huawei’s ascend 910C chips in the inference processes provides a critical advantage for Chinese companies. This Arad Huawei can further strengthen his position in this field. The company is being discussed that the Nvidia has developed a new generation Ascend AI chips that will rival the Blackwell series. If this happens, it is inevitable that competition in the global artificial intelligence equipment market will become further.