Nvidia persuaded U.S. export regulators that restricting its shipments to China would hand an opening to Huawei, and Washington appears to have taken that warning seriously. Reporting by The Wire China reveals Nvidia supplied U.S. officials with an internal assessment projecting that Huawei had already sold hundreds of thousands of AI chips and could scale to roughly a million shipments in 2025. Combined with other evidence the company presented, that case helped tilt the outcome in Nvidia’s favor.
The episode crystallizes a persistent tension in the U.S.-China technology standoff: export controls are meant to slow Beijing’s advance in sensitive areas, but enforcing those controls can impose high costs on U.S. firms and create incentives for Chinese suppliers to accelerate indigenization. Nvidia’s lobbying shows how those commercial and strategic pressures intersect, and why Washington’s choices are far from straightforward.
What Nvidia reportedly told U.S. officials
According to The Wire China, Nvidia delivered a written report to U.S. government decision‑makers estimating Huawei sold between 200,000 and 450,000 AI chips in 2024 and could reach about 1 million in 2025. Nvidia argued that tighter U.S. restrictions would not halt Chinese progress but instead open a market vacuum Huawei could fill with domestically produced accelerators.
Nvidia CEO Jensen Huang also raised the point directly with lawmakers, warning that export curbs risked accelerating Huawei’s competitiveness rather than containing it. In Huang’s telling, Nvidia’s own commercial footprint in China had effectively evaporated — he stated Nvidia’s share in China had “dropped to zero.” The company further disclosed that it had written off roughly $5.5 billion in H20 chips developed specifically for the Chinese market.
The company additionally provided a performance comparison: Huawei’s 910C accelerator delivers roughly one‑third of the performance per chip compared with Nvidia’s Blackwell generation. That gap, Nvidia argued, could be overcome by scale — if Chinese customers lost access to U.S. accelerators at the same time Huawei scaled production, Huawei could capture substantial market share despite the per‑chip deficit.
The Wire China reports that those arguments contributed to the U.S. decision to authorize additional Nvidia shipments to China.
Why the argument cut through
Nvidia presented more than commercial harm; it framed the issue as a classic second‑order effect of export policy. U.S. controls aim to slow China’s access to high‑end compute for defense and AI capabilities. But if restrictions simply cause U.S. suppliers to exit the market, Chinese customers will buy from domestic vendors — and domestic vendors stand to gain scale, revenue, and learning that shorten the time to parity.
For U.S. policymakers, that tradeoff is morally and strategically fraught. A strict embargo can blunt immediate access to advanced chips, but it also cedes market share and manufacturing experience to competitors. By quantifying the scale of Huawei’s existing shipments and projecting future volumes, Nvidia made a concrete case that U.S. action could accelerate the very outcome export controls were intended to prevent.
Nvidia had its own motives. The company’s write‑off of approximately $5.5 billion in H20 chips developed for China is a tangible cost of losing access to that market. Reopening channels preserves revenue and protects long lead investments in products tailored for Chinese customers. That self‑interest does not negate the geopolitical logic, but it does frame the company’s presentation to policymakers: remaining commercially engaged in China can, in Nvidia’s view, reduce long‑term strategic risk.
Technical gaps don’t erase commercial impact
Nvidia’s comparison, that Huawei’s 910C performs at roughly one‑third the level of Blackwell on a per‑chip basis, highlights another key point: performance supremacy is necessary but not sufficient to guarantee dominance. Scale, integration, software ecosystems, and customer relationships matter. A domestic supplier that ships millions of moderately capable accelerators can enable the widespread deployment of AI services and models tailored to local needs, and those deployments, in turn, generate data, applications, and talent advantages.
Nvidia’s numbers suggest it judged Huawei was already moving quickly, and that removing U.S. competition in the short term would give Huawei an outsized growth runway. Whether that projection proves accurate will be an important test of export control theory: will temporary access denials delay capability growth, or will they accelerate domestic alternatives that bring China closer to parity?
Broader implications for U.S. export policy
This episode underscores a structural problem for Washington. Export controls are a blunt instrument applied in a globalized supply chain where commercial incentives shape the diffusion of technology. When U.S. firms successfully persuade regulators that sales restrictions would produce undesirable strategic consequences, authorities face an uncomfortable choice: uphold a strict interpretation of security goals or carve exceptions that preserve market access but risk undercutting those same goals.
The debate is not hypothetical. Analysts at the Peterson Institute for International Economics have argued that ad‑hoc model and export controls can backfire and unintentionally bolster Chinese capability building. The idea that export policy must be both strategically coherent and operationally enforceable is gaining traction in policy circles as regulators calibrate measures meant to be both deterrent and durable.
Nvidia’s intervention is a vivid example of industry influence shaping that calibration.
What to watch next
Two metrics will be decisive. First, independent verification of Huawei’s shipment volumes and the performance trajectory of domestically produced accelerators will determine whether Nvidia’s warning was prescient. Second, U.S. export policy moves in coming months, will regulators adopt clearer, more durable rules that balance security and commercial concerns, or will approvals and carve‑outs continue on a case‑by‑case basis?
For further context on the policy fight and the firms jockeying in it, see EastFrontier’s investigation of banned Nvidia servers that exposed export-control gaps, our reporting on Huawei’s Ascend 910C and market demand, and coverage of Jensen Huang’s criticism of export controls.
Nvidia was ultimately persuasive enough to secure more sales in China. Whether that was the prudent strategic choice, or a tactical retreat that speeds Chinese self‑sufficiency, is an open question that will shape the next phase of the U.S.-China chip competition.