Speaking Wednesday at the opening ceremony of a supply chain expo in Beijing, Huang described Chinese-developed AI models as “world class,” citing examples such as DeepSeek, Alibaba, Tencent, MiniMax, and Baidu’s Ernie bot. He credited these systems with helping to drive AI innovation well beyond China’s borders.
“Models like DeepSeek, Alibaba, Tencent, MiniMax, and Baidu Ernie bot are world class, developed here and shared openly [and] have spurred AI developments worldwide,” Huang told attendees. He also highlighted Nvidia’s entrenched presence in China’s technology sector, noting that more than 1.5 million local developers use Nvidia’s tools and platforms to bring their ideas to life.
His visit to Beijing marks his third trip to China this year, underscoring the company’s efforts to maintain its position in a critical but increasingly complicated market. Nvidia, the dominant global supplier of AI training chips, has faced mounting regulatory hurdles over the last three years as the U.S. government imposed successive rounds of export controls on advanced semiconductors out of security concerns.
Those restrictions, aimed at limiting China’s access to technology that could strengthen its defense industry, have taken a toll on Nvidia’s bottom line. Huang acknowledged in May that the U.S. curbs had nearly halved Nvidia’s market share in China, costing it $2.5 billion in lost sales during the April quarter alone. The company has warned of an additional $8 billion hit in the July quarter, projecting quarterly sales at $45 billion.
On Tuesday, however, Nvidia announced it expected to soon resume shipments of its H20 AI chip to China, following new assurances from the U.S. government that provided clarity on export rules. Sales of the chip had been on hold since April due to tightened requirements.
Among the companies Huang singled out for praise was DeepSeek, a startup that startled global investors in January with an AI model offering lower development and operating costs than OpenAI’s. While it’s unclear exactly how DeepSeek built its model under U.S. chip restrictions, reports suggest its parent company High-Flyer stockpiled Nvidia hardware before curbs tightened.
Huang also emphasized China’s embrace of open-source AI development, contrasting it with more restrictive approaches seen in the West. “China’s open-source AI is a catalyst for global progress, giving every country and industry a chance to join the AI revolution,” he said. He described open-source technology as essential for AI safety and for fostering international cooperation on standards.
Recent releases by Chinese firms underscore this approach. Last week, Alibaba-backed Moonshot AI released a new open-source model called Kimi K2, which claims to outperform OpenAI’s ChatGPT and Anthropic’s Claude on certain coding benchmarks.
Huang also pointed to the integration of AI into daily life in China, describing how it powers popular apps like Tencent’s WeChat, Alibaba’s Taobao, ByteDance’s Douyin, and Meituan’s delivery service.
Meanwhile, geopolitical dynamics around technology trade continue to evolve. Last month’s U.S.-China trade talks in London resulted in tentative steps toward easing frictions, with the U.S. beginning to loosen some export restrictions on advanced technologies and China resuming certain rare earth export licenses.
While Nvidia and other U.S. firms continue to face scrutiny over potential military applications of their technology in China, Huang has argued that maintaining global access is vital if the U.S. wants to stay at the forefront of AI innovation. In a recent interview with CNN, he pushed back on fears that sales to China would automatically bolster its military capabilities, framing broad technology sharing as essential for global progress in AI.