Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Niphon | Istock | Gets the image
Favorable and BakeThe two largest technology companies in China have shown how they are holding in the global race of artificial intelligence, even when the United States is pulling curbs on key semiconductors.
Business methods include chips that make AI models more effective and even using homegrown semiconductors.
While US President Donald Trump administration The rock is one contradicted Biden’s chip ruleIt’s still Exports of some semiconductors from companies including Nvidia and Amd In April.
Big names in the sector considered the issue during the last call calls.
Martin Lau, President Tencent – Operator of China’s largest messaging app – stated that his company has “fairly strong reserves” of the chips he had purchased before. It meant blocks of graphic processing (graphic processors), a type of semiconductor that became a gold standard for the preparation of huge AI models.
These models require a powerful computational transmission provided by graphic processors to process high data volumes.
But, Lau said, contrary to the opinion of US companies that the GPU clusters should expand to create a more advanced AI, Tencent can achieve good learning results with a smaller group of such chips.
“It actually helped us look at the existing high-quality chips inventory and say that we should have a high-end enough to continue training models that go forward,” Lau said.
As for the conclusion – the process of actual fulfillment of the AI task, not just training – Lau said that Tencent uses “software optimization” to improve efficiency to expand the same number of graphic processors to perform a particular function.
Lau added that the company is also looking for the use of smaller models that do not require such high computing power. Tencent also said it could use the chips and semiconductors that are currently available in China.
“I think there are many ways (B) that we can meet the expanded and growing needs, and we just need to continue to study these places and spend more time on software, not just the gross strength of buying graphic processes,” Lau said.
The approach of kaida
Baido, the largest search campaign in China, has advertised what he calls his capabilities of “full state”-the complication of its cloud computing infrastructure, AI models and actual applications based on these models, such as its Ernie Chatbot.
“Even without access to the most advanced chips, our unique full opportunities AI AI allow us to create strong applications and provide considerable value,” said Dun, President AI Cloud Business Baidu, this week.
Baidu also advertised software optimization and the ability to reduce the cost of launching its models because it has most of the technology in this stack. The Baidu guide also spoke about the efficiency that allows it to get more from the graphic processors he owns.
“With the help of foundations that excite the need for large -scale computing power, the ability to build and manage GPU large -scale clusters and effectively use graphic processors became key competitive advantages,” Shen said.
The Baidu leader has also flared up to the progress of domestic Chinese technology firms in AI Semiconductors.
“Internally designed self-sufficient chips, as well as (an) an increasingly effective home software stack, jointly creates a strong basis for long-term innovations in the Chinese ecosystem AI,” Shen said.
Focus on the internal chip in China
Over the past few years, China has been increasing the development of chips developed and manufactured on their home soil. Most experts agree that Beijing remains overall for the US in the field of graphic processes and ah, but there were some success.
Gavrov Hupta, an analyst covering semiconductors in Gartner, said the stock is one way that Chinese companies are engaged in export restrictions. In addition, China has made some progress in semiconductor technology, even if it remains for the United States, Gupto added.
“China is also developing its own internal semiconductor ecosystem, but only from materials to equipment to chips and packaging. Different segments have made different levels of progress, but China was amazingly consistent and ambitious for this purpose, and we must admit that they have achieved decent success,” Gupto CNBC said.
“It gives them avenue to buy AI chips that may not compete with those who are US chips but continue to progress.”
Many US leaders called on Washington to lift export restrictions in light of China’s progress. CEO Nvidia Jensen Huang Named the “Failure” curb This week, they say they do more harm to US business than China.
