Tencent Says It Has Enough AI Chip Stockpile for Models and Many Options ami...

　　TMTPOST -- Tencent Holdings Ltd. on Wednesday floated enough artificial intelligence chip stockpile and many alternatives, playing down the potential impact of pause in supply of Nvidia Corporations H20 chips.

　　Credit：Tencent

　　“We do have enough chips for training and continuous upgrade of our existing models,” Tencent President Martin Lau told analysts at an earnings call on Wednesday. Lau also said Tencent has “many options” when it comes to the chips for running AI models, without specifying what kind of chips on its list of potential alternatives.

　　Lau’s remarks came as Beijing is becoming more skeptical on the security of Nvidia’s H20, a scaled-down AI chip that Nvidia had specially tailored to the Chinese market to comply with the Biden-era AI chip export controls.

　　Nvidia CEO Jensen Huang in mid-July said the U.S. government had approved the resumption of H20 chip sales to China. But China’s Cyberspace Administration on July 31 called Nvidia representatives into a meeting to discuss what it deemed serious security vulnerabilities with the H20 chip. The agency asked staff to explain potential risks and provide documents as needed, clouding the outlook of H20 sales.

　　U.S. President Donald Trump on Monday confirmed reports that Nvidia will be allowed to sell H20 chips to China, in exchange for giving the U.S. government 15% of the revenue they get from the sales.

　　Tencent, as another Chinese tech titan focusing on AI technology and accelerating spending on AI research, still heavily relies on Nvidia chips for the training of its AI models, The Information cited the China’s AI industry insider on Wednesday.

　　Lau on the earnings conference Wednesday said the status of imports of AI chips from U.S. was unclear yet as Beijing and Washington continued negotiations, We dont really have a definitive answer on the import situation yet. There are a lot of discussions between the two governments. We are waiting to see what exactly comes out of that, said Lau.

　　Lau implied a more measured approach to AI development than many other rivals as Tencent seeks sustainable monetization from its AI initiatives. “We also have to be spending smartly, rather than just saying we’re going to go all-in and spend on buying a lot of chips, and sort of hiring a lot of people and doing a lot of marketing,” said Lau, adding that Tencent will “spend in the right tempo.”