In a development that reveals the complex interplay between technological ambition and geopolitical reality, Nvidia is restarting shipments of its H200 AI chips to China after receiving multiple U.S. government approvals and purchase orders from Chinese customers. The move comes as CEO Jensen Huang makes bold predictions about the AI hardware market reaching $1 trillion in revenue over the next two years, while simultaneously navigating volatile trade policies and delayed diplomatic summits.
The China Deal: A Delicate Balancing Act
Nvidia’s journey back to the Chinese market has been anything but straightforward. After pausing production earlier this year due to regulatory uncertainty, the company has now received licenses for “many” Chinese customers and is restarting manufacturing of the H200 chips. These chips represent a generation behind Nvidia’s current products, reflecting the compromise struck in December’s deal with the White House that allows sales while giving the U.S. a 25% cut of income.
“We’ve been licensed for many customers in China for the H200. We have received purchase orders from many customers. And we are in the process of restarting our manufacturing,” Huang told reporters at Nvidia’s annual GTC conference in San Jose. The situation, he noted, was “different than it was two weeks ago… our supply chain is getting fired up.”
The $1 Trillion Vision
While navigating export restrictions, Huang has simultaneously painted a staggering picture of AI hardware’s future. At the same GTC event, he predicted that AI hardware revenue will reach at least $1 trillion over roughly the next two years, driven by rapid adoption of AI agents and increasing demand for computing power. This forecast significantly exceeds Wall Street consensus estimates of about $835 billion for Nvidia’s 2027-2028 revenue.
Huang’s vision extends beyond immediate market access to long-term technological dominance. He unveiled the new Groq 3 language processing unit chip, designed to speed up AI system responses, which will be manufactured by Samsung and shipped in Q3 2026. More dramatically, he projected $1 trillion in orders for Nvidia’s Blackwell and Vera Rubin chips through 2027 – a doubling from last year’s $500 billion demand forecast through 2026.
Geopolitical Headwinds and Market Realities
The timing of Nvidia’s China market re-entry coincides with broader geopolitical complications. U.S. President Donald Trump has requested to delay a high-stakes summit with Chinese President Xi Jinping by about a month due to the ongoing Iran war, which has disrupted global oil supply and threatened U.S. prices. This summit was expected to address AI technology exports among other issues.
Trump’s break with historically tight U.S. export controls on advanced semiconductors has prompted pushback from national security hawks in Washington. Huang addressed this tension directly: “President Trump’s intention is that US should have a leadership position and access to Nvidia’s best technology. However, he would also like us to compete worldwide and not concede those markets unnecessarily.”
Technological Evolution and Competitive Landscape
Looking beyond immediate market access, Nvidia is already planning its next technological leaps. The company’s upcoming Feynman AI accelerator, scheduled for 2028, will feature 3D-stacked GPU dies and custom High-Bandwidth Memory (cHBM), representing a significant advancement in chip design. This technology addresses heat dissipation challenges with potential power consumption over 2000 watts, positioning Feynman as the successor to Rubin and Rubin Ultra accelerators.
Meanwhile, the AI industry faces growing scrutiny over ethical implementation. A recent lawsuit against Elon Musk’s xAI highlights concerns about AI models producing abusive content, with plaintiffs alleging that the company’s Grok model created sexual images of minors without basic precautions. This case underscores the regulatory and ethical challenges that accompany rapid AI advancement.
Market Implications and Business Considerations
For businesses and professionals, Nvidia’s China re-entry represents both opportunity and complexity. Chinese tech giants like Alibaba and ByteDance could soon regain access to Nvidia’s AI technology, potentially accelerating China’s AI development. Huang has estimated that China’s AI chip market could be worth up to $50 billion, offering a significant revenue stream for the $4.5 trillion chip company.
However, investors remain cautious. Despite Huang’s bullish forecasts, Nvidia’s stock briefly surged then fell back, reflecting ongoing concerns about AI investment returns, supply chain threats from Middle East conflicts, and memory chip shortages. The company’s shift to Samsung for Groq 3 manufacturing, departing from typical TSMC use, adds another layer of supply chain complexity.
As AI hardware evolves from Blackwell to Rubin to Feynman architectures – with Rubin operating 3.5x faster than Blackwell on model-training tasks and 5x faster on inference tasks – businesses must navigate not just technological capabilities but also geopolitical permissions, ethical considerations, and market volatility. The path forward requires balancing innovation with regulation, global competition with national security, and technological potential with practical implementation.