Chinese Huawei runs its capabilities in the chips industry

Digest opened free editor

Rola Khaleda, FT editor, chooses her favorite stories in this weekly newsletter.

Huawei founder said that the United States is exaggerating the capabilities of the Chinese chip maker and the decline in its company’s technology amid commercial talks between Beijing and Washington that includes discussions on export controls.

In a rare interview with Daily, the Chinese state people on Tuesday, Ren Chengvy said HuaweiChip Ascend Chip, the main competitor of NVIDIA products in China, “still falls behind by the United States by one generation.” He added that “the United States has exaggerated the capabilities of Huawei – we are not so yet.”

Ren’s comments are CEO of Nvidia Jensen Huang The alarm appeared In recent weeks about Huawei’s progress with artificial intelligence chips, saying that Washington’s borders on the American chip maker who sells to China have led to a “massive” rival threatening to undermine American domination of artificial intelligence technology.

The United States and China started a new round of commercial talks on Monday in London, which included discussing Washington’s export controls on major technology.

During the first round of the talks in Geneva, the United States did not discuss export controls. However, Beijing’s latest restrictions on some rare lands and minerals used to manufacture cars – threatening to close factories lines in the United States, Europe and Japan – brought the issue to discussions in employment.

Huawei benefited from Washington’s ban on NVIDIA chips to China, where Chinese technology giants accelerate the ascension chips and Make preparations To adopt Huawei technology.

However, the majority of Chinese AI companies, including Deepseek, use NVIDIA chips to train large language models that operate artificial intelligence tools. Local alternatives are increasingly used for the less complex task of calling models to create responses in tools such as Chatbots.

Analysts and researchers in Huawei previously did He complained about the technical defects In using the company’s chips for training, quoting difficulty making chips work together and distributing computing work burdens through them.

REN hinted on Tuesday that the company has made great steps to solve these problems, saying that Huawei could “make up” the poorest performance through cluster computing, which includes linking multiple chips to increased capacitance of artificial intelligence servant’s power.

He said: “Using assembly and stacking, our computing results are similar to the best in the world.”

The new Cloudmatrix 384 Ai Server is the new Huawei Axis of its strategy to compete with NVIDIA by collecting a large number of chips to support the domain capacity and deal with more data. The server connects 384 AI processors using Huawei optical technology.

The person participating in the test said that many Huawei customers are already testing the Cloudmatrix server, and they work with the company’s engineers to solve the heat problems resulting from the large number of chips that work in complete harmony and weighing the machine.

The person added that Huawei’s biggest challenge is to create an “environmental system” for developers who will use her platform to build artificial intelligence models. One of the advantages of NVIDIA is the Cuda software platform, which developers say it is easy to use.

Ren said that Huawei was investing 180 billion yuan (25 billion dollars) annually on research and development, with 60 billion yuan entry into basic research that is not intended to develop products, but they are conducting penetration discoveries.

He added that China has distinctive advantages when building its technological capabilities.

He said: “Artificial intelligence depends on abundant electricity and infrastructure of the advanced network.” “The world -class power and network generation systems are the most advanced in the world.”

Additional reports by Dimitri Sevastopolo in Washington