The tectonic plates of the global AI industry are shifting under Nvidia's feet. Chinese companies are abandoning the American chip giant's advanced accelerators at a pace that surprised even industry watchers, according to a new Bloomberg Intelligence survey that captures the real-world impact of US-China tech decoupling.
The numbers that signal a structural shift
Executives at China's largest AI infrastructure builders say they will allocate 46% of their AI accelerator budgets to domestic products within the next 12 months. That is a sharp jump from the current 30% — and it represents billions of dollars flowing away from Nvidia toward Chinese semiconductor suppliers like Huawei, according to the survey released Tuesday.
Why Chinese companies are walking away from Nvidia
This is not a gradual preference shift. It is a forced realignment driven by US export controls that have progressively restricted Nvidia's ability to sell its most advanced chips — including the H100, H200, and Blackwell series — to Chinese buyers. For Chinese tech giants building massive AI data centers, relying on American chips has become a strategic liability.
Who is leading the domestic chip pivot
China's biggest AI infrastructure builders — Tencent Holdings Ltd., Alibaba Group Holding Ltd., and Huawei Technologies Co. — are at the center of this transition. These companies are not just buyers; Huawei is also a key domestic supplier, creating a unique dynamic where one of China's largest tech firms competes with Nvidia while also powering its rivals' AI ambitions.
The cost pressure behind the shift
The survey revealed that 80% of executives reported their total AI infrastructure spending is running over-budget this year. The primary reason: the exceptionally high cost of AI-related projects. This budget pressure adds another incentive for Chinese firms to explore domestic alternatives that may offer lower costs or more predictable supply chains.
How US export controls reshaped the market
The Biden administration and subsequent US governments have progressively tightened restrictions on advanced semiconductor exports to China, citing national security concerns. These controls specifically target Nvidia's most powerful AI accelerators. The result has been the opposite of what US policymakers may have intended: rather than slowing China's AI ambitions, the restrictions have supercharged Beijing's push for domestic chip self-sufficiency.
What this means for Nvidia's business
Nvidia has long relied on China for a significant portion of its revenue. While the company has developed lower-spec versions of its chips specifically for the Chinese market to comply with export rules, the survey suggests these workarounds may not be enough. Chinese companies appear to be making a strategic bet on domestic suppliers for the long term, not just a temporary workaround.
Confirmed facts vs what remains unclear
Confirmed: The Bloomberg Intelligence survey data showing 46% domestic allocation target and 80% over-budget spending. The involvement of Tencent, Alibaba, and Huawei as key players. The role of US export controls in driving the shift.
Unclear: Which specific domestic suppliers will capture the increased spending. Whether Chinese domestic chips can match Nvidia's performance in advanced AI workloads. The exact timeline of the transition. Whether this trend will accelerate further if US restrictions tighten.
Huawei's dual role: competitor and supplier
Huawei's position in this story is unique. The company is both a major buyer of AI accelerators for its own cloud and AI services, and a supplier of domestic alternatives through its Ascend chip series. This vertical integration gives Huawei significant influence over China's AI infrastructure direction, though it also creates potential conflicts of interest with other Chinese tech firms that compete with Huawei.
Risks and balanced view of the domestic shift
The transition to domestic chips carries significant risks. Chinese semiconductor suppliers still lag behind Nvidia in raw performance, software ecosystem maturity, and developer tools. The CUDA platform that Nvidia has built over years gives it a massive advantage in ease of use and developer familiarity. Chinese alternatives may require significant software re-engineering and may not match Nvidia's performance for the most demanding AI training workloads.
The broader pattern: global semiconductor realignment
This story is part of a larger trend reshaping the global semiconductor industry. Countries from the US to Europe to Japan are investing heavily in domestic chip production capacity. China's push is the most aggressive, driven by both national security concerns and industrial policy. The AI accelerator market, once dominated by a single company, is fragmenting along geopolitical lines.
What this means for investors and tech buyers
For investors, the message is clear: Nvidia's China revenue is structurally at risk, and Chinese semiconductor suppliers represent a growing opportunity. For technology buyers globally, the fragmentation of the AI chip market means more options but also more complexity. Companies may need to maintain relationships with multiple suppliers to hedge against geopolitical disruptions.
What happens next in the China-Nvidia story
The trajectory depends heavily on US policy. If export controls tighten further, the shift to domestic Chinese chips will accelerate. If restrictions ease, some Chinese companies may return to Nvidia — but the survey suggests the domestic ecosystem has already gained enough momentum that a full reversal is unlikely. Chinese firms are building long-term relationships with domestic suppliers, not just making temporary adjustments.
Our Take
This survey captures a genuine inflection point. The shift from 30% to 46% domestic allocation in just 12 months is not incremental — it is structural. What makes this story significant is not just the numbers but the strategic intent behind them. Chinese companies are not being forced to buy domestic chips because of immediate shortages; they are choosing to invest in domestic alternatives as a long-term strategic hedge. That choice, once made, is hard to reverse because it involves building software ecosystems, training engineers, and establishing supply chain relationships. Nvidia may have lost China not because its chips are inferior, but because geopolitics made them unreliable.
Frequently Asked Questions
Why are Chinese companies leaving Nvidia?
Chinese companies are shifting to domestic AI accelerators primarily because of US export controls that restrict Nvidia from selling its most advanced chips to China. The survey shows this is a strategic, long-term move rather than a temporary adjustment.
Which Chinese companies are leading the shift away from Nvidia?
Tencent Holdings, Alibaba Group, and Huawei Technologies are the key players driving this transition. Huawei is notable because it is both a buyer of AI accelerators and a supplier of domestic alternatives through its Ascend chip series.
How much are Chinese companies spending on domestic AI chips?
According to the Bloomberg Intelligence survey, Chinese executives plan to allocate 46% of their AI accelerator budgets to domestic products within the next 12 months, up from 30% currently. 80% of executives also reported that total AI infrastructure spending is running over-budget this year.
Can Chinese domestic chips replace Nvidia's performance?
Chinese domestic chips currently lag behind Nvidia in raw performance and software ecosystem maturity, particularly for the most demanding AI training workloads. However, the gap is narrowing, and for many applications, domestic alternatives may be sufficient. The transition requires significant software re-engineering.
