Nvidia expects its projected $200 billion market for central processing units to include China, CEO Jensen Huang said on arrival in Taipei on Saturday, reaffirming the company’s view of long-term demand despite persistent U.S.-China technology frictions.
Huang first outlined the $200 billion opportunity during an earnings call Wednesday, tying the potential to a broadening market for CPUs as enterprises adopt agentic AI - systems designed to carry out autonomous tasks - and expand their spend beyond graphics processing units used to train large AI models.
When asked by reporters whether the $200 billion estimate encompassed China, Huang responded simply: "I would think so." He reiterated the strategic significance and scale of the Chinese market, while acknowledging that regulatory and licensing matters affect how Nvidia can serve customers there.
Licensing and approvals for Nvidia’s H200 chips remain a puzzle. The H200 - described by company executives as Nvidia’s second-most-powerful AI chip - has received licenses from the U.S. government to be sold into China, but Chinese regulatory approval has not yet been granted. Separately, while it has been reported that the U.S. has cleared around 10 Chinese firms to acquire H200 hardware, Huang noted that no deliveries have taken place so far.
Huang also referenced recent high-level diplomacy. Talks in Beijing this month between U.S. President Donald Trump and Chinese President Xi Jinping produced no immediate breakthrough that would enable wider shipments of H200 chips, he said, and noted he was in Beijing as part of the U.S. delegation.
On the product front, Huang returned to the company’s vision for a CPU-led addressable market and highlighted Nvidia’s new line of central processors, named Vera. He said Vera processors open the door to that larger CPU market opportunity and form part of a broader product roadmap that Nvidia expects to help sustain growth beyond its core AI accelerator business.
Huang described the company’s Vera Rubin platform - which pairs the Vera CPU architecture with the Rubin GPU architecture - as entering an accelerated production ramp. He said the combination will make for a "very busy second half" for Taiwan’s semiconductor supply chain as Nvidia scales manufacturing and integration of those platforms.
Huang is in Taipei ahead of next month’s Computex trade show and said he plans to meet with TSMC, the world’s largest contract chipmaker and a key manufacturing partner for many advanced semiconductors. He added that Nvidia has invested in and supported partners in Taiwan and suggested the company’s involvement there has exceeded public announcements.
The comments came amid wider industry activity in Taiwan. AMD announced on Thursday an investment exceeding $10 billion to deepen partnerships and expand capacity to build and assemble advanced AI chips in Taiwan, a move Huang referenced while speaking about supply-chain dynamics in the region.
Separately, Taiwanese prosecutors said they are investigating three people suspected of illegally exporting high-end AI servers built by Super Micro that contained Nvidia chips subject to U.S. export controls. The matter has drawn scrutiny because of the potential for diversion of controlled technology.
Asked what further steps Nvidia could take to limit diversion risks, Huang said the company is "very rigorous" in educating partners about applicable laws and regulations and expects partners to comply. "Ultimately, Super Micro has to run their own company," he said, adding: "I hope that they will enhance and improve their regulation compliance and avoid that from happening in the future."
In a related legal development, the U.S. Justice Department in March charged three people associated with Super Micro, including a co-founder, with allegedly aiding the smuggling of at least $2.5 billion of U.S. AI technology to China in violation of export rules.
Summary
Jensen Huang told reporters in Taipei that Nvidia’s $200 billion CPU market forecast includes China. While U.S. licenses for the H200 AI chip exist, Chinese approval and actual deliveries remain pending. Nvidia is ramping its Vera Rubin platform and will engage Taiwan’s supply chain, even as authorities investigate possible illegal exports involving Super Micro servers with Nvidia chips.
Key points
- Nvidia says its $200 billion CPU market estimate includes China, signaling expected demand across key territories.
- H200 chips have U.S. licenses but await Chinese approval and no deliveries to China have occurred yet despite reported clearances for some firms.
- Nvidia is increasing production of its Vera Rubin platform and plans engagements with Taiwan supply-chain partners, amid broader industry investments in Taiwan.
Risks and uncertainties
- Regulatory approvals and export controls could limit Nvidia’s ability to ship H200 chips to China - impacting semiconductor sales and global AI hardware supply chains.
- Investigations and criminal charges related to alleged smuggling of AI servers introduce compliance and legal risks for vendors and downstream buyers in the server and enterprise hardware sectors.
- Supply-chain capacity and ramp timing in Taiwan will affect product availability for platforms like Vera Rubin and could influence production schedules for AI deployments.