Nvidia CEO Jensen Huang delivers the keynote address during the Nvidia GTC 2025 at the SAP Center in San Jose, California on March 18, 2025. | Justin Sullivan—Getty Images
Nvidia announced on Tuesday that it will record a $5.5 billion charge in its upcoming quarterly earnings—one of its largest ever—due to unsold H20 graphics processing units (GPUs) that were meant for China and other restricted markets. The news sent Nvidia shares diving more than 6% in extended trading and triggered broader selloffs across the semiconductor sector.
This charge stems directly from new export license requirements imposed by the U.S. Department of Commerce on April 9, which now limit Nvidia’s ability to sell these AI chips to China, Saudi Arabia, the UAE, and several other countries. The impact is massive—not just for Nvidia, but for the entire AI hardware industry, which has heavily relied on global demand.
The H20 GPU, a specialized AI chip built for Chinese markets, was designed to comply with U.S. export restrictions introduced in 2022 and revised again in 2023. Despite those efforts, Washington has tightened the screws even further, claiming that such processors could potentially fuel military-grade AI systems or supercomputers.
According to internal estimates, Nvidia was expecting between $12 billion to $15 billion in 2024 revenue from H20 sales alone. That figure now hangs in limbo.
“This charge marks the most visible sign yet that U.S. export policy is hitting Nvidia’s growth engine hard,” said Gene Munster, managing partner at Deepwater Asset Management.
The aftershocks weren’t limited to Nvidia. Competitor Advanced Micro Devices (AMD) saw its stock drop 7% after hours, while Broadcom fell nearly 4%. With Nvidia being one of the top-performing stocks of the AI boom, its downturn could signal a cooling off in the red-hot chip market.
To put it in perspective:
Investors are now grappling with how much longer these regulatory restrictions will persist—and what other sectors might be impacted.
While the U.S. remains Nvidia’s largest market, China ranked fourth by revenue in Nvidia’s latest annual report, behind the U.S., Singapore, and Taiwan. In 2023, China accounted for nearly $10.3 billion in revenue, a sharp drop from pre-restriction years. CEO Jensen Huang revealed in a February earnings call that Chinese revenue had been slashed by more than 50%.
Moreover, Nvidia recently highlighted Huawei as a growing competitor in the AI hardware space. This is the second consecutive year Huawei has appeared in Nvidia’s annual threat disclosures—a clear signal that local Chinese players are catching up.
Adding to the pressure, DeepSeek, a rising Chinese AI research firm that recently launched its R1 AI model, had been using Nvidia’s H20 chips in training before these new bans took effect.
The H20 chip is based on Nvidia’s 2022 Hopper architecture—the same foundation as the widely popular H100 and H200 chips used in Western markets. However, the H20 version was intentionally throttled with slower memory bandwidth and interconnect speeds to stay under U.S. export thresholds.
It was a workaround—but now, even that workaround is being pulled.
“We’re running out of room to innovate for the global market while complying with export rules,” Huang warned during Nvidia’s GTC 2024 conference.
With the H20 sidelined, Nvidia is shifting focus to its Blackwell architecture, unveiled earlier this year. These chips are significantly more powerful and will cater primarily to Western markets and enterprise clients in AI research, cloud computing, and autonomous systems.
However, export challenges remain. The Biden administration has also introduced new “AI diffusion rules,” set to take effect next month. These rules could place further restrictions on the distribution of U.S.-based AI technologies and tighten control over overseas collaboration.
Interestingly, Huang also emphasized that “half the world’s AI researchers are Chinese”, many of whom work in U.S.-based labs. This underscores the complex interdependence between the two superpowers in tech innovation. Despite geopolitical tension, collaboration remains vital—but may now become riskier or more politically charged.
Nvidia has reportedly moved its testing, validation, and distribution facilities out of China in response to previous export bans, signaling its longer-term pivot away from risky regions.
In addition to chip-specific regulations, tariff threats are re-emerging under former President Donald Trump’s new policy proposals. On April 2, Trump announced sweeping new tariffs targeting America’s top trading partners—including China and the EU. While some exemptions were granted for tech products like smartphones and semiconductors, they may be temporary.
The implication? Nvidia and other U.S. chipmakers could face double pressure from both export restrictions and import taxes, eroding their global competitiveness even further.
With first-quarter results scheduled for May 28, Nvidia is now facing heightened scrutiny. Between the $5.5 billion write-down, looming regulatory changes, and increased competition from Chinese firms like Huawei, the company may have to rethink its global roadmap.
Analyst consensus remains cautiously optimistic, with some maintaining a “Buy” rating due to Nvidia’s leadership in AI computing. But expectations for near-term volatility are rising fast.
“This isn’t just a business story—it’s a geopolitical chess match,” said Wedbush analyst Dan Ives. “And Nvidia is caught in the middle.”
Nvidia's setback is more than a financial charge—it’s a turning point in the tech industry’s global expansion story. As nations clamp down on strategic technologies and the AI race heats up, the lines between innovation, regulation, and diplomacy are becoming increasingly blurred.
Investors, regulators, and innovators alike should brace for more turbulence—and more headlines like this.