Nvidia Faces Scrutiny Over AI Chip Exports Amid Industry Optimism

Nvidia US67066G1040

Nvidia finds itself at the center of renewed political scrutiny in Washington. A U.S. House of Representatives committee is examining whether the chipmaker’s technical support for the Chinese AI startup DeepSeek may have indirectly facilitated military applications in China. This congressional inquiry coincides with an ongoing debate over export controls for AI semiconductors, even as positive signals emerge from the industry's supply chain.

According to Reuters, John Moolenaar, a Republican representative from Michigan and chair of the House Select Committee on the Chinese Communist Party, sent a letter to U.S. Commerce Secretary Howard Lutnick on January 28. The letter is based on Nvidia’s internal documents.

These records reportedly show that Nvidia’s technical development staff assisted DeepSeek in training its AI models. The correspondence describes an "optimized co-development of algorithms, frameworks, and hardware."

The cited documents indicate that the DeepSeek-V3 model required 2.788 million H800 GPU hours for full training. The letter notes this figure is below what U.S. developers typically use for cutting-edge AI models.

Moolenaar concedes that at the time of the support in 2024, there was no public indication DeepSeek's technology could be used by China's military. Consequently, Nvidia treated DeepSeek as a legitimate commercial partner and provided "standard technical support."

Company Rebuttal and the H800 Chip

Nvidia has rejected the allegation that it is making China militarily dependent on U.S. chips. The company states that China possesses "more than enough domestic chips for all military applications, with millions in reserve." It argues that the U.S. military using Chinese technology is as implausible as the Chinese military relying on U.S. technology.

The investigation focuses on the H800 chip. This model was specifically developed and sold for the Chinese market before the U.S. implemented export controls on certain high-performance chips in 2023.

Broader Policy Debate on Export Controls

The heightened congressional attention occurs during a tense debate over AI chip exports to China. In early January, the Trump administration essentially approved the sale of Nvidia's more powerful H200 chips to China, albeit with conditions. Among these, deliveries to companies supporting the Chinese military are prohibited.

This decision has drawn bipartisan criticism from China hawks in Washington. During a January 15 hearing, Matt Pottinger, a former senior Asia advisor in the White House during Trump's first term, stated the administration is "on the wrong track" regarding AI policy.

This increases political pressure on the entire industry to establish clear boundaries concerning military-adjacent applications in China. Nvidia is central to this scrutiny because its graphics processing units are considered key components for training large AI models.

Contrasting Positive Signals from a Key Supplier

In contrast to the political headwinds, more favorable signals are emerging from the industry side. ASML, the Dutch manufacturer of chipmaking equipment and a crucial supplier for producers of Nvidia's chips, reported record orders this week.

For the fourth quarter of 2025, net bookings reached 13.2 billion euros, significantly surpassing analyst expectations of 6.32 billion euros cited by CNBC. For 2026, ASML forecasts revenue between 34 and 39 billion euros, with the midpoint of this range above market estimates.

CFO Roger Dassen attributed the strong demand to customers' "more robust view" on the sustainability of the AI boom. This assessment has prompted some buyers to expand their capacity and deliberately ramp up medium-term production.

For Nvidia, this implies that despite political risks, the structural demand outlook for high-performance AI chips remains intact, at least from the perspective of equipment suppliers.

Upcoming Catalyst: Quarterly Earnings Report

The next concrete insight into Nvidia's business performance will come in late February. According to its investor relations calendar, the release of figures for the fourth quarter of fiscal year 2026 is scheduled for February 25.

The company is expected not only to provide fresh data on demand for its AI chips but also to comment on the impact of recent and potential future export restrictions. For the market, growth in the data center segment and any hints of adjustments to its China strategy will be particularly crucial.