China has announced it will allow top artificial intelligence companies to purchase a limited number of H200 chips from Nvidia Corp., marking a significant easing of restrictions on the coveted US technology, according to The Information. Chinese officials have informed companies such as Alibaba Group Holding Ltd., ByteDance Ltd. and DeepSeek that they will have permission to purchase some of the processors, which are essential for developing AI models. The companies must provide detailed information about their chip requirements and justify their needs to receive approval for purchases, as reported by The Information.
Chinese authorities have held meetings with leading technology companies over the past month to discuss potentially restricting overseas access to China's most advanced AI models, including those yet to be released, according to Reuters sources. The discussions, led by China's Ministry of Commerce, reflect Beijing's growing effort to treat cutting-edge AI as a strategic national asset, similar to the United States, amid intensifying global competition over the technology. Companies that attended the meetings included Alibaba, ByteDance and startup Z.ai, with participants discussing imposing restrictions on the country's most advanced AI models, including both closed-source and more open versions. Officials also discussed making the leak or theft of proprietary AI technology an offence under China's stringent national security law, with the scope of proposed measures still under consideration. The discussions also covered strengthening legal protections for AI technology, with officials raising the possibility of introducing new measures to restrict who can invest in domestic AI startups.
In June, the US Commerce Department's Bureau of Industry and Security ordered Anthropic to suspend access to its two most advanced models — Claude Fable 5 and Claude Mythos 5 — for certain foreign nationals after officials raised concerns about a possible technique that could bypass the models' cybersecurity safeguards. The restrictions were later lifted after Anthropic addressed the government's concerns, restoring global access on July 1. As legal publication Lawfare observed, this marked the first known instance of the US applying export controls to an AI model itself rather than to the semiconductors or computing infrastructure used to build it. Until recently, governments sought to influence AI development by controlling the inputs required to create it — advanced chips, chipmaking equipment and high-end compute. The Anthropic episode suggested that governments were now willing to control the finished capability itself, establishing an important precedent for global AI governance.
For India, the immediate concern is not whether the US or China is justified in restricting access to frontier AI models, but whether such access can still be taken for granted. Over the past two years, Indian startups and enterprises have built products using models from both ecosystems — OpenAI, Anthropic, Google and Meta alongside China's Qwen and DeepSeek — selecting whichever offered the best balance of capability, cost and performance. Thomas J Vallianeth, partner at Trilegal, told Business Standard: "The lesson of the past month is that model access is now a supply-chain risk. Boards will begin to treat AI dependency accordingly and diversify the regions from where foundational models are procured." Ashish Tandon, founder and CEO of Indusface, argued that recent developments have fundamentally changed the conversation around sovereign AI, stating that "sovereign capability has been an India ambition for the long term; this reframes it as near-term resilience policy."
Chinese authorities have held meetings with leading technology companies over the past month to discuss potentially restricting overseas access to China's most advanced AI models, including those yet to be released, according to Reuters sources. The talks, led by China's Ministry of Commerce, reflect Beijing's growing effort to treat cutting-edge AI as a strategic national asset, similar to the United States, amid intensifying global competition over the technology. Companies that attended the meetings included Alibaba, ByteDance and startup Z.ai, with participants discussing imposing restrictions on the country's most advanced AI models, including both closed-source and more open versions. Officials also discussed making the leak or theft of proprietary AI technology an offence under China's stringent national security law, with the scope of proposed measures still under consideration. The discussions also covered strengthening legal protections for AI technology, with officials raising the possibility of introducing new measures to restrict who can invest in domestic AI startups.
Chinese AI models have gained significant traction worldwide with the release of DeepSeek's R1 model last year because of their affordability and growing capabilities. Any move by Beijing to restrict access to these products could have implications for global AI markets, potentially increasing costs for many businesses. According to Reuters, two of the sources told the publication that Chinese authorities are extremely concerned that Washington may use the model against Chinese interests and that Mythos might exploit software weaknesses. This aligns with concerns expressed in public by state media and Zhou Hongyi, the founder of 360, a cybersecurity company that serves government and business clients. Hongyi has stated that China must create its own mythology. The extent of the possible limitations is still up for debate and might only be applicable to models that are developed in the future, with it not immediately clear when or even if they would come into force.
