Image source: unsplash
Nvidia Corp. used the opening of CES 2026 to project confidence about both its technology roadmap and its ability to navigate one of its most sensitive geopolitical challenges: the future of its business in China.
At a keynote address in Las Vegas on Tuesday, Nvidia's CEO Jensen Huang unveiled advances in what the company calls “agentic AI” and “physical AI,” while confirming that Nvidia’s next-generation Rubin architecture has entered full-scale production, with products slated for volume shipment in the second half of the year.
Rubin succeeds the Blackwell platform and is positioned as the company’s flagship system for large-scale AI training and real-time inference across robotics, industrial automation and digital agents.
Behind the technical announcements, however, Nvidia’s remarks about China drew equal attention from investors and analysts.
At a press conference later that day, Huang said the company had received US government approval to export its H200 artificial intelligence chips and had begun ramping up production to meet demand.
“Market demand is very strong,” Huang said. “We’ve already activated our supply chain.”
In a notable shift from his more cautious tone over the past year, Huang added: “It looks like we’ll be returning to China.”
Still, he warned that no formal announcements should be expected and that even internally, visibility would remain limited until actual purchase orders arrive and export licenses are granted.
The comments follow a decision announced by U.S. President Donald Trump on Dec. 8 to ease restrictions on Nvidia’s AI chip exports to China, allowing the company to apply for licenses to sell the H200 — a processor positioned below Nvidia’s most advanced Blackwell and Rubin chips.
The move triggered a wave of inquiries from Chinese customers, according to people familiar with the matter, though whether and when shipments will be approved remains uncertain. Reuters reported late last year that Nvidia was aiming to ship H200 chips to China ahead of the Lunar New Year, and that it had already planned for higher production capacity.
Nvidia said at the time that selling H200 to authorized Chinese customers would not compromise its ability to serve clients elsewhere.
China once represented one of Nvidia’s largest growth engines. In fiscal 2024, the company generated $17.1 billion in revenue from the Chinese market, a 66% year-on-year increase that accounted for 13.1% of total sales. Huang previously suggested the market could ultimately grow to $50 billion annually.
That outlook collapsed in 2025 after Washington tightened controls on advanced chip exports. Nvidia’s market share in China’s AI accelerator segment fell from about 95% to near zero within a single quarter, according to industry estimates, and the company removed China from its forward revenue guidance.
The situation was further complicated by security scrutiny around Nvidia’s earlier H20 chip, which became entangled in debate over potential “backdoor” risks, dampening prospects for near-term recovery.
Nvidia has been careful to stress that its long-term financial projections do not rely on China. At a press briefing, Huang said the company’s forecast of $500 billion in revenue over the next five quarters — issued last year — excludes any contribution from potential H200 sales to China. He also reiterated that Washington would not permit exports of the more advanced Blackwell or Rubin processors to Chinese customers.
During a fireside chat with analysts, Stacy Rasgon of Bernstein questioned whether Nvidia could realistically balance supply and demand if Chinese orders suddenly materialized, and whether the supply chain could remain stable amid shifting regulatory conditions.
Chief Financial Officer Colette Kress said demand in China was “clearly there,” but emphasized that Nvidia would not divert supply from existing customers in the US and other markets to serve China.
“We’re not going to reallocate capacity that’s already committed,” she said, adding that the company remains dependent on the US government’s export approval process.
Kress described the easing of restrictions as “the best opportunity for us to compete fairly on a global scale and to offer China a truly outstanding product,” while noting that customers have already filed license applications and that regulators are actively reviewing them.
She also highlighted the strategic importance of the Chinese market.
“China is a very large economy with world-class engineers and AI talent,” she said. “If we are able to sell under any future licenses, its growth trajectory could eventually resemble what we see in the US.”
For Nvidia, the stakes go beyond revenue. Maintaining a presence in China helps preserve ecosystem influence, software adoption and long-term relevance in one of the world’s most important AI development hubs — even if near-term volumes are limited.
At CES, Nvidia’s technological narrative remained firmly focused on scale, performance and integration. Huang framed Rubin as the foundation for a new phase of AI in which models move beyond text and images into real-world perception, robotics and autonomous systems.
That vision resonated with investors, helping keep Nvidia at the center of the CES spotlight through a packed schedule of press briefings, analyst meetings and partner events.
Yet the company’s ability to execute that strategy globally still depends on navigating political boundaries as carefully as technical ones.
As Huang acknowledged, the company’s return to China — if it happens — will be incremental, regulated and uncertain. For now, Nvidia is betting that even a partial reopening is better than none, while its growth engine remains anchored in markets where policy, demand and supply are more predictable.
“You should be happy to hear that demand is strong,” Huang said, smiling beside his CFO. But for Nvidia, strength in demand is only part of the equation. Approval, timing and geopolitics will decide whether that demand can turn into sales.
