Nvidia will no longer sell its H200 chips in China, ending the company's participation in a controversial loophole in the U.S. ban on advanced chip exports — and the announcement, delivered on the company's first earnings call since Jensen Huang's resignation, came the same day The Wall Street Journal reported that the Securities and Exchange Commission has opened a civil probe into Nvidia's disclosures.
The twin developments mark the sharpest policy pivot yet for the chipmaker since Huang stepped down last week, and they close the book on the export arrangement that defined — and increasingly haunted — the company's China business. For the full picture of the earnings numbers behind the announcements, see the latest AI industry coverage alongside Nvidia's reported results.
The End of the H200 Loophole
Acting CEO Colette Kress confirmed the China decision on Tuesday. Nvidia's deal to sell H200 chips into China was brokered by Huang and President Donald Trump, and the arrangement required the chips to be shipped through the United States — allowing the U.S. government to collect a 25 percent tariff on each sale.
The pathway angered China hawks in Washington who wanted to deny Chinese labs access to leading AI accelerators, but Huang long argued that competing in China was strategically essential. In April, he told podcaster Dwarkesh Patel that "to concede that market for the United States technology industry is a disservice to our country."
That argument is now moot inside his own company. Huang's ouster came in part over revelations of an affair with the CEO of a company under investigation for smuggling chips into China, and his departure has coincided with a hardening of U.S. policy: new rules issued Monday bar advanced U.S. chips from foreign data centers that serve Chinese customers remotely.
New Rules, New Board Committee
The China announcement came the same day Nvidia said it was creating a board-level committee to oversee national security policies, export control compliance and global trade risks — an institutional signal that the company is restructuring how it manages geopolitical exposure.
Investment bank Jefferies has forecast that the new remote-access rules could trim between $44 billion and $62 billion from Nvidia's data center revenue next year. Kress, however, said the backlog of demand for advanced chips such as the upcoming Vera Rubin generation means the restrictions will have no impact on the company's top-line revenue.
The SEC Probe
According to The Wall Street Journal, the SEC is opening a civil investigation into whether the company's disclosures misled investors. The probe adds to a mounting stack of legal and political exposure: Huang resigned following reports of extramarital affairs, including one with Alice Huang, an executive at Singapore-based data company Megaspeed. The Commerce Department has been investigating Megaspeed since late 2024 on suspicion that the company diverted Nvidia chips to China in violation of U.S. export laws.
Congress has demanded that the former CEO answer questions about the affair and how it might have affected his public stance that export restrictions should be loosened. Treasury Secretary Scott Bessent put the criticism more bluntly over the weekend, accusing Nvidia of knowingly selling powerful chips to China to "goose their revenues and their earnings."
Nvidia did not immediately respond to requests for comment on the reported probe, which remains civil — a step short of enforcement action, but an escalation from congressional letters and public criticism.
China's Countermove
Even before Tuesday's announcement, Chinese customers had largely stopped accepting H200 deliveries. Beijing has preferred to generate demand for its domestic chip industry, even though major Chinese AI compute users still want Nvidia's superior hardware, Jacob Feldgoise, a researcher at Georgetown's Center for Security and Emerging Technology, told TechCrunch.
Meanwhile, China's frontier labs have found ways to access Nvidia silicon through other routes. All of China's leading AI models show evidence of having been trained on Nvidia Grace Blackwell chips that are prohibited from being exported to China — training that may have taken place at data centers outside the country. Under the new rules, large data centers suited for AI training that want to buy Nvidia chips will have to certify they are not serving any Chinese customers, export control attorney Dan Fisher-Owens told TechCrunch. Facilities that cannot provide that certification may be pushed toward alternative chips, further eating into Nvidia's market share.
A Structural Shift for the AI Chip Market
Analysts have long argued that Nvidia's China-linked demand is understated in official disclosures: annual revenue at its Singapore subsidiary hit $104 billion in 2025 before the company stopped disclosing those figures, and critics claim as much as 20 percent of Nvidia's revenue is tied to demand connected to China. That exposure is now being dismantled by policy, by Beijing's own procurement preferences, and by Nvidia's decision to walk away.
For Chinese AI labs, the retreat accelerates the push toward domestic accelerators and alternative supply chains. For U.S. allies hosting large data centers, the new certification regime turns every major GPU purchase into a geopolitical attestation. And for Nvidia itself, the episode transforms what was once a growth channel into a compliance perimeter — governed, from this week forward, by a dedicated board committee.
---
Stay Informed on AI PolicyExport controls, regulation, and the AI industry's biggest stories — covered daily.
Read more AI news →