Nvidia Corp Chief Executive Jensen Huang delivered a definitive assessment of the company’s diminishing position in China during a live broadcast on CNBC, signaling a sharp reversal from the dominant market standing the American semiconductor giant held only three years ago.
Appearing in his signature leather jacket, Huang stated that the outside world should no longer harbor optimistic expectations regarding the import of Nvidia's top-tier graphics processing units (GPUs) into the Chinese market. Acknowledging that the silicon pioneer has lost its competitive edge within the region due to escalating geopolitical barriers, Huang took the aggressive operational step of completely removing projected revenues from China from the corporation’s upcoming earnings guidance—a move analysts view as a structural admission of defeat.
The operational shift marks a severe downturn from 2023, when Nvidia controlled an estimated 95 percent of China’s lucrative artificial intelligence (AI) accelerator market. The contraction is heavily linked to successive rounds of tightening export controls by the U.S. Department of Commerce, which effectively blocked Nvidia from shipping its flagship architectures. Compelled to offer heavily downgraded, high-priced regional variants with compromised performance—such as the H20 chip—Nvidia inadvertently incentivized Chinese enterprise consumers to transition to domestic alternatives.
The Domestic Avalanche
Industry data from 2025 illustrates the rapid erosion of Nvidia’s market share. Total cumulative shipments of AI accelerator cards in China reached four million units, with local semiconductor brands supplying 1.65 million units to capture over 40 percent of the market. Conversely, Nvidia’s regional shipping volume plummeted to 2.2 million units, dragging its market share down to 55 percent.
Independent real-world benchmarking has further accelerated the transition. Technical evaluations reveal that the inference performance of Huawei Technologies' Ascend 950PR accelerator card operates at 2.87 times the speed of Nvidia's compliance-tailored H20 chip. Facing a choice between restricted, artificial foreign components and fully functional domestic flagship hardware, major Chinese hyperscalers have shifted their procurement strategies based strictly on commercial metrics.
Industry sources confirm that ByteDance has placed an order for 400,000 Huawei Ascend 950PR chips in a transaction valued between 20 billion and 30 billion yuan ($2.76 billion to $4.14 billion), while technology conglomerates Alibaba and Tencent are reportedly in queues for similar allocations. Huawei aims to mass-produce 750,000 Ascend 950PR units this year, though current supply chain capacity remains severely bottlenecked by overwhelming domestic demand.
Systemic Supply Chain Shift
The historical corporate preference for imported silicon has been largely dismantled by structural supply chain security risks. Chinese technology firms are increasingly favoring computing platforms that offer stable operation, competitive price-to-performance ratios, and complete independence from external Western regulatory constraints.
Alternative domestic architectures are rapidly closing the training gap. The Hygon Deep Computing 2 processor reportedly achieves a training efficiency equivalent to 80 percent of Nvidia’s legacy A100 chip, utilizing a CUDA-compatible architecture that drastically lowers software migration barriers for developers. Concurrently, Cambricon’s MSI 370 card has gained traction by targeting the mid-tier market, offering a cost-performance advantage priced at one-third of Nvidia’s A10 GPU.
Wall Street financial institutions have swiftly adjusted their long-term valuation models to reflect the geopolitical reality:
Bernstein: The research firm projects a pessimistic outlook, predicting Nvidia's market share in China will contract to just 8 percent by the end of this year, while local Chinese brands are expected to command over 60 percent of the domestic market.
Morgan Stanley: The banking institution issued a long-term forecast estimating that China’s domestic AI chip market will swell to $67 billion by 2030, with domestic semiconductor enterprises controlling more than 70 percent of that total valuation.
Geopolitical Counter-Effects
The systemic containment strategy orchestrated by Washington has generated unintended secondary effects in secondary markets. Following Western sanctions that blocked Russia’s state-owned Sberbank from procuring Nvidia hardware, Russian enterprises systematically pivoted toward Huawei’s computing ecosystem, accelerating the formation of non-Western supply networks.
Market analysts note that the rapid decline of a global leader from a 95 percent monopoly to a single-digit market projection serves as a stark reminder of how geopolitical restrictions can backfire on domestic industries. By forcing Chinese technology firms to prioritize internal R&D and supply chain resilience, Washington's export controls have effectively transformed local backup systems into primary enterprise choices, fundamentally rewriting the rules of global technological competition.
No comments:
Post a Comment