H100 image [NVIDIA]
Huang argued that the restrictions had not only dampened the firm’s revenue but spurred Chinese firms to become more competitive in AI. Specifically, blocking U.S. chips forced Chinese buyers to switch, primarily, to Huawei-class silicon and pour capital into a domestic supply chain the curbs were meant to constrain, he said.
In the long run, strengthening the Chinese market could threaten the ability of the U.S. to carve out a sustainable lead in AI, even after Chinese firms like DeepSeek have already substantially leveled the playing field with powerful yet inexpensive offerings such as the R1 reasoning model. “More than half the world’s AI researchers are based in China … Our competition there is really intense. They’d love for us never to go back,” said Huang, per Reuters.
China is ascendent in AI hardware and software
Huang later estimated the total revenue “damage” from the H20 export blockage could hit $15 billion. That figure would cut roughly $10.9 billion in gross profit, based on Nvidia’s 72.7% FY2024 gross margin. To put that in context, the lost profit alone is about 1.26× the company’s entire $8.68 billion FY 2024 R&D budget.
Competition is also intense on the GPU front, with Nvidia and Huawei’s offerings increasingly going head-to-head in the Chinese market with Nvidia’s market share in China dipping from 95% to 50%.
| Timeframe | U.S. (Nvidia) GPU developments | Chinese (Huawei) GPU developments | Notes |
|---|---|---|---|
| 2019 | Ascend 910 launched (August 2019) | Huawei’s early entry into high-performance AI training chips. | |
| 2020 | Nvidia A100 released (May 2020) | Nvidia’s flagship data center GPU, setting a new performance benchmark. | |
| 2022 | Nvidia H100 announced (March 2022), shipping late 2022. October 2022: Significant U.S. export controls imposed, restricting A100/H100 sales to China. |
Ascend 910B reportedly debuted by some sources; positioned as comparable to A100. | The initial major wave of U.S. restrictions begins to reshape the market. |
| 2023 | Nvidia develops China-compliant variants like A800/H800. October 2023: U.S. export controls further tightened, impacting A800/H800 and future designs like the H20. |
Ascend 910B gains traction; Baidu reportedly places significant orders (orders around August 2023). | Chinese firms actively seek alternatives owing to to Nvidia chip restrictions. Huawei’s Ascend 910B emerges as a domestic option. |
| Early-Mid 2024 | Nvidia H20 (next-gen China-compliant flagship) faces shipment delays and market challenges following tightened U.S. export controls. Nvidia announces a $5.5 billion quarterly charge (April 2025 related to Q1, but reflecting impact from H20 issues in 2024). | Tightened controls make it increasingly challenging for Nvidia to supply even modified high-end chips to China. | |
| Mid-Late 2024 | Ascend 910C reportedly “came into existence” (August 2024), with testing and orders. Positioned as an alternative to Nvidia’s H20 and aiming to rival the H100. | Development of more powerful domestic alternatives in China continues. | |
| 2025 | Nvidia continues to navigate export controls for its next-gen GPUs. Nvidia CEO Huang estimates total revenue “damage” from H20 export blockage could hit $15 billion. | Ascend 910C mass shipments anticipated May 2025. Ascend 910D samples for select Chinese firms expected (late May 2025), mass production planned Q4 2025. Designed to rival/surpass Nvidia’s H100. |
The competitive landscape intensifies. |
A balancing act
The pursuit of an R&D presence in Shanghai, even with Nvidia’s assurance that no core GPU designs would be modified there, point to a long-term plan to assess local customer needs and counter rising domestic competition. But at the same time, Nvidia must maneuver continued Washington-imposed constraints on the Chinese market, which are easing in some markets like United Arab Emirates (UAE) and India, but not in China.
In recent years, prominent U.S. chipmakers have reported substantial hits to their China-related business. Nvidia, for example, disclosed that the October 2022 export rules initially put at risk about $400 million in quarterly sales of its A100 GPUs to China. In response, Nvidia designed the A800 and H800 chips to comply with the rules. However, further tightening in 2023–24 threatened its next-generation H20 flagship. In April 2025, Nvidia announced it would incur a $5.5 billion charge in its fiscal first quarter owing to the inability to ship H20 processors to China. The company also projected an additional $15 billion in lost revenue from the Chinese market as a result of these restrictions.
AMD has also adjusted its forecasts for similar reasons. The company expects to forgo $1.5 billion in 2025 revenue because its MI300 AI accelerators (MI308 variants) are now subject to licensing requirements. AMD could incur up to an $800 million charge related to inventory and purchase commitments tied to these restricted chips.
Intel hasn’t disclosed a specific dollar impact, but with approximately 29% of its $53 billion annual revenue in 2024 derived from China, the opportunity cost is significant if Chinese customers, from PC makers to data centers, pivot away. Intel has also announced significant layoffs as part of its cost-reduction measures. In April 2025, the company confirmed plans to cut approximately 15,000 jobs, following a previous reduction of 15,000 positions in late 2024. These workforce reductions are aimed at streamlining operations and reducing expenses.
