Key Takeaways
- New Trump administration regulations would mandate US approval for artificial intelligence chip exports to nearly all nations, marking a significant expansion beyond current limits.
- Orders exceeding 1,000 Nvidia GB300 GPUs would trigger government review processes; massive deployments of 200,000+ units need host nation approval.
- Production of H200 chips intended for China has ceased at TSMC, with manufacturing resources redirected toward Vera Rubin processors.
- CFO Colette Kress disclosed that Nvidia hasn’t earned any revenue from Chinese markets, even with authorization for certain H200 deliveries.
- Jensen Huang indicated Nvidia’s $30 billion stake in OpenAI’s funding round might be final before anticipated public offering.
Shares of Nvidia $NVDA declined approximately 1.7% on Thursday following a one-two punch of unfavorable developments that rattled investor confidence.
According to Bloomberg’s reporting, the Trump White House is formulating unprecedented export regulations requiring federal authorization for AI semiconductor sales to practically every nation globally. The news pushed NVDA alongside $AMD, which fell roughly 2%, into negative territory during afternoon session.
These proposed regulations would transform current limitations — presently affecting approximately 40 nations — into a comprehensive worldwide licensing system. The framework would subject any order of up to 1,000 GB300 GPUs from Nvidia to governmental evaluation, though certain exemption pathways may exist.
Bulk purchases face heightened examination. Any deployment surpassing 200,000 GB300 processors controlled by a single entity within one nation would necessitate involvement from that country’s government in the authorization procedure.
Washington would exclusively approve such substantial exports to partner nations willing to provide security guarantees and commit capital toward American AI development — although the proposal lacks concrete investment threshold details.
While not constituting an outright prohibition, these measures would grant the Commerce Department extensive authority over access to semiconductors powering platforms like ChatGPT and Gemini.
Chinese Market Remains Frozen
In related news, Financial Times disclosed that Nvidia has discreetly halted H200 chip manufacturing for Chinese customers at Taiwan Semiconductor Manufacturing Co., reallocating production resources to its forthcoming Vera Rubin processor line.
These chip families employ distinct technologies and fabrication processes — H200 utilizes CoWoS-S packaging combined with previous-generation high-bandwidth memory, whereas Vera Rubin incorporates CoWoS-L packaging alongside cutting-edge HBM4 specifications — meaning the production reallocation doesn’t directly impact availability of either product line.
Nvidia’s Chinese operations have remained stagnant for several months. While Trump greenlit H200 distribution to China last December, the authorization came with requirements for a 25% government revenue share. Previously, Nvidia marketed the less powerful H20 processor in China — until the administration prohibited those sales last April.
Despite receiving authorization, transactions haven’t materialized. During last week’s quarterly results presentation, CFO Colette Kress acknowledged Nvidia has “yet to generate any revenue” from Chinese territories and remains uncertain whether Beijing will permit any imports whatsoever.
Domestic Chinese Competition Intensifies
Kress highlighted additional challenges: multiple recent public offerings from Chinese semiconductor firms that she stated “have the potential to disrupt the structure of the global AI industry over the long term.” Nvidia indicated it will maintain dialogue with authorities in both Washington and Beijing.
Regarding OpenAI developments, CEO Jensen Huang revealed this week that Nvidia’s $30 billion commitment to OpenAI’s $110 billion February financing round “might be the last time” the semiconductor manufacturer invests in the artificial intelligence venture, anticipating OpenAI’s imminent transition to public markets. Huang further noted that a previously discussed $100 billion investment arrangement with OpenAI is “not in the cards.”
