N Nvidia has broken its silence over its stalled push to re-enter China's AI chip market.

Nvidia has officially broken months of silence over its stalled push to re-enter China's AI chip market. At the company's annual GTC developer conference in San Jose, California, CEO Jensen Huang confirmed that both the US and China have cleared the path for H200 GPU sales, and that production lines are being switched back on. "We have received purchase orders from many customers" Huang told reporters at GTC, "We've been licensed for many customers in China for H200. We have received purchase orders from many customers, and we're in the process of restarting our manufacturing." He added that the situation was markedly different from just weeks ago. "That's new news for all of you, and it's different than it was two weeks ago or three weeks ago, but that's our condition today," Huang said. The announcement marks a turning point for Nvidia, which had seen zero revenue from China's data centre market despite months of negotiations and regulatory approvals. China clears the final hurdle Sources confirmed to Reuters that Beijing has now issued licences for many customers in China, following US approvals that had already come through. A spokesperson for the Chinese embassy in Washington said they were "not aware of the specifics" and directed questions to "the competent authorities." Huang's remarks suggest Beijing has eased its earlier pressure on local buyers, and that restarting Hopper-class silicon manufacturing for China once again makes commercial sense for Nvidia. Why was Nvidia unable to sell to China? The path here was anything but smooth. Prior export controls had forced Nvidia to develop a lower-capability chip for Chinese markets called the H20. After President Donald Trump initially halted those sales, he reversed course in December and allowed Nvidia to ship the more advanced H200 to China, provided the US received a 25 percent cut of sales. The Commerce Department formally published the licensing framework in January 2026, with Trump confirming, "We're allowing them to do it, but the United States is getting 25 percent of the chips, in terms of the dollar value." The approval comes with strict oversight. China-bound sales are capped at no more than 50 percent of the total number of H200 chips sold to US customers, and Nvidia must certify that sufficient stock is available for the domestic US market. But as recently as late February, Nvidia's CFO Colette Kress had warned investors that the company's Q1 2027 revenue forecast did not account for a single cent of revenue from Chinese data centre customers. What is the H200 AI chip? The H200 sits below Nvidia's current Blackwell architecture but remains roughly six times more powerful than the H20 - the downgraded chip Nvidia had originally designed to stay within earlier export limits. It is what Chinese hyperscalers have been waiting for. China will prohibit the use of these chips in government agencies, military applications, critical infrastructure, or state-owned enterprises. Sales are restricted to commercial and technology customers. Demand has been significant, reports state, with orders from Chinese technology companies have reportedly totalled more than 2 million H200 chips, far exceeding Nvidia's current inventory of about 700,000 units, with each chip priced at around $27,000. Blackwell and Rubin remain off-limits Nvidia's more advanced chips are not part of this deal. Nvidia's top-of-the-range Blackwell and forthcoming Rubin series remain banned for sale in China and were not included in the H200 agreement. Huang noted that his forecast of over $1 trillion in revenue by 2027 reflects sales of Blackwell and Rubin systems alone - suggesting significant upside from H200 China sales on top of that.

Physicists at the University of Oxford show they can assemble Schrödinger’s cat-like superpositions from far more exotic quantum ingredients.

Selection marks the next phase of construction on Micron’s New York project, expected to generate 50,000 jobs.

OneAdvanced has launched what it bills as the UK’s first private, sovereign healthcare large language model trained on NHS primary-care data.