Deepseek gives China chip manufacturers to the cheapest running

The rise of Deepseek artificial intelligence models is providing some Chinese chip producers, such as Huawei, a better chance to compete in the internal market against the most powerful processors produced in the United States.

For years, Huawei and its Chinese colleagues have fought to combine Nvidia in the development of cutting -edge chips that could compete with the products of the US company in the AI ​​training models, a process in which the data are provided to algorithms to help learn make accurate decisions.

However, the Deepseek models, which focus on inference or when an artificial intelligence model produces conclusions, optimize computational efficiency rather than depend only on gross processing power.

This is one of the reasons why the Chinese start model should partially decrease the difference between what artificial intelligence processors produced in China and its most powerful US equivalent can do, analysts say.

Huawei and other chip producers to the Chinese, such as Hygon, encams, supported by Tencent; Tsingmicro and Moore thread have made statements in recent weeks stating that the products will support the Deepseek models, although few details have been released.

Huawei did not comment on the topic. Moore Threads, Hygon Encame and Tsingmicro did not answer Reuters’ questions.

The managers of the sector are now predicting that the Open Source nature of Deepsek and the low price charged by the company to use the model can increase the adoption of the AI ​​and the development of real technological applications, helping Chinese companies to overcome the restrictions from exports imposed by the United States.

Even before Deepek wins this year’s titles, products such as Huawei ASCEND 910B have been seen by customers as a bytedance as more suitable for less intensive tasks of “inference”, the internship after training in which the IA models have trained do Forecasts or perform tasks, such as chatbots, such as chatbots, such as chatbots, such as chatbots, such as chatbots.

In China, dozens of companies from car manufacturers to telecommunications suppliers, have announced plans to integrate the Deepsek models with their products and operations.

“This development is very aligned with the ability of chipset suppliers,” said Lian Jye su, analyst of the Omdia technological research company.

“The Chinese chipsets of AI have difficulty competing with the NVIDIA GPU (graphic processing unit) in the training of the AI, but the workloads of the Inference of the AI ​​are much more tolerant and require a much more local and specific knowledge of the sector.”

Nvidia still dominates

However, Bernstein’s analyst, Lin Qingyuan, said that although the Chinese chips are competitive for inference, this is limited to the Chinese market, since the Nvidia chips are even better, also for inference activities.

Although the restrictions of US export prohibit the entry of the most advanced training chips in Nvidia in China, the company is still authorized to sell less powerful training chips that Chinese customers can use for inference activities.

Nvidia published an article on Thursday on how the inference time is increasing and has argued that its chips will be necessary to make Deepsek and other most useful “reasoning” models.

In addition to the calculation power, the Cudda of Nvidia, a parallel processing platform that allows software developers to use the company’s GPUs for general use, not only for artificial intelligence or graphic intelligence, has become a crucial component of theirs domain.

Huawei was the most aggressive in his efforts to break with Nvidia, offering a Cuda equivalent called Calcol Architecture for Neural Networks (Cann), but the experts said they had faced obstacles to persuade developers to abandon the North Company-American platform platform .

“The performance of the software of Chinese AI chip companies are also insufficient at this stage. Cuda has a rich library and a diversified range of software resources, which requires a long -term significant investment,” said Sudia.

Source: Terra

Rose

Rose James is a Gossipify movie and series reviewer known for her in-depth analysis and unique perspective on the latest releases. With a background in film studies, she provides engaging and informative reviews, and keeps readers up to date with industry trends and emerging talents.