Nvidia caught between U.S. export controls and China's pushback as AI chip war intensifies

U.S. Approves Nvidia H200 Chips to China as Policy Shifts

The United States has approved exports of Nvidia's H200 artificial intelligence chips to China, marking a sharp reversal in Washington's tech policy 4

4

. This decision comes after Nvidia CEO Jensen Huang spent much of last week meeting with U.S. legislators, including President Trump and Republican members of the Senate Banking Committee 1

1

. The move represents a significant shift for Nvidia, which was expected to sell approximately $17 billion worth of chips to China this year before restrictions brought that figure down to zero 4

4

. Current analyst estimates include nothing for China sales, making this approval an incremental positive, though it comes with considerable uncertainty around tariffs and market access.

China Limits Access Despite U.S. Approval

Beijing is signaling it will limit access to the advanced AI chips, requiring permits and explanations for purchases 4

4

. Chinese authorities have spent months pushing back on the on-again, off-again availability of Nvidia hardware by encouraging domestic companies to use domestic chip suppliers where possible, mandating that Chinese companies use at least 50% domestically produced hardware 1

1

. Beijing has quietly added domestic AI chips from firms including Huawei to an official government procurement list for the first time, a move intended to expand the use of homegrown semiconductors across government agencies and state-owned firms that could generate billions of dollars in orders for local chip designers 5

5

. Chinese chipmaking stocks extended losses, with Hua Hong Semiconductor declining 2.6% and SMIC sliding 2% in Hong Kong trading 5

5

.

Senate SAFE Bill Threatens 30-Month Export Ban

Despite the recent approval, the proposed Secure and Feasible Exports Act (SAFE) bill would force the Commerce Department to halt export licenses for the sale of the latest chips to U.S. adversaries, including China and Russia, for 30 months 1

1

. This ban could cover all existing chips and anything more powerful developed by major companies over that period. Although it primarily targets Nvidia's Blackwell GPUs, it would also cover Nvidia's last-generation Hopper designs, AMD's graphics chips, and Google's latest TPU designs 1

1

. This represents devastating news for Nvidia and many chip-manufacturing contemporaries, as China remains a massive market for hardware and AI development.

Military Modernization Concerns Drive Export Controls

Analysis of dozens of procurement documents published by the People's Liberation Army reveals that the Chinese military is directly soliciting and using advanced U.S. chips, including those designed by Nvidia, to develop AI-enabled military capabilities 3

3

. One contract for an "intelligent optoelectronic target recognition system" specifies the use of Nvidia computing resources, while another notice for server procurement to help the Chinese military "perform AI algorithm calculations" relies on Nvidia H100 GPUs, a chip that was export controlled in 2022 3

3

. The PLA is requesting systems that can generate, collect and analyze troves of battlefield data to quickly identify targets and accelerate decision-making cycles, with other documents featuring requests for algorithms to power swarms of autonomous vehicles 3

3

. Authoritative Chinese documents indicate that Beijing believes the development and deployment of advanced AI-enabled military systems provides the best chance to catch up to or surpass the U.S. military 3

3

.

Training Workloads Still Depend on Nvidia Hardware

Nvidia's GPU versatility is particularly well-suited for AI training, and it has no real rival in this domain 1

1

. When DeepSeek developers were forced to use locally produced chips for training, they ended up switching back to Nvidia hardware because the performance just wasn't there 1

1

. Frontier model training still largely relies on Nvidia hardware, and Chinese access to cutting-edge Nvidia chips will make it easier for frontier AI labs to acquire advanced computing power to train increasingly capable AI models 3

3

. Nvidia chips are very strong for general-purpose computing, and anytime you need Nvidia's CUDA software, there is really no better substitute 4

4

.

China Advances Domestic Chip Alternatives Despite Efficiency Gaps

China has made major leaps in its AI hardware development over the past year as it sought to build more reliable access to powerful AI hardware while the U.S. turned the tap on and off 1

1

. These conditions have led Huawei to make tremendous advances and design high-power systems that scale well, at the expense of efficiency 1

1

. Recently announced Huawei SuperClusters are more powerful than any Nvidia system, despite not using the most advanced AI chips 2

2

. China claims that new chip packaging and assembly techniques can close the performance gap between Nvidia and its local producers 1

1

. Chinese chip firms have announced enormous plans to manufacture several times the chips they managed in 2025 as soon as next year 1

1

.

Geopolitical Frictions Reshape Semiconductor Supply Chain

U.S. export controls have made this an issue of national pride and led to a wave of investment into a domestic AI chip ecosystem within China 2

2

. It's unclear if the U.S. will ever regain market share even if chip controls are reversed, as China has also retaliated in many ways that have further hurt the U.S. economy and geopolitics . The U.S. has lost what could have been one of the largest markets for advanced AI chip companies . Despite all the blocks and barriers from various governments and organizations, it hasn't been too difficult for companies to allegedly get their hands on Nvidia hardware, with Singaporean companies allegedly used to circumvent trade blocks, and leasing computing power from international partners effectively allowing Chinese national companies to use whatever hardware they like 1

1

. The way to win the AI war is with continued innovation rather than restricting sales, as staying ahead on the innovation front matters more than limiting market access 4

4

.