NextFin News - Chinese companies are shifting more of their AI hardware budgets away from Nvidia and toward domestic suppliers, a sign that export controls and Beijing’s self-reliance push are reshaping one of the world’s most important technology supply chains. In a Bloomberg Intelligence survey released Tuesday, executives in China said they expect 46% of their AI-accelerator budgets to go to domestic products over the next 12 months, up from 30% today. The same survey found 80% of executives said their infrastructure spending is running over budget this year, underscoring how expensive the country’s AI buildout has become.
The figures matter because they point to a market shift that is no longer only about access to Nvidia’s most advanced chips. It is also about procurement patterns, software adaptation, system integration, and the practical cost of building AI infrastructure under restrictions and local industrial policy. Chinese firms are not simply turning away from a single supplier; they are reorganizing their hardware stack around domestic alternatives that can be bought, deployed, and supported inside a politically constrained market.
The result is a direct challenge to Nvidia’s China franchise. The company remains one of the dominant suppliers of AI accelerators globally, but the Chinese market has been narrowing as buyers test local options and as policymakers continue to encourage homegrown technology. That creates two layers of pressure at once: weaker addressable demand for Nvidia in China and a stronger incentive for Chinese vendors to accelerate the development of domestic chips, networking gear, cooling, and software tooling.
The survey also fits into a broader national investment cycle. China is preparing to spend around 2 trillion yuan, or about $295 billion, over five years on a nationwide data-center buildout, with state firms expected to operate much of the infrastructure and local suppliers favored for at least 80% of the technology, including AI chips. That plan, outlined in June, shows the scale of the industrial policy behind the current buying shift. The issue is not just whether Chinese companies can technically obtain Nvidia hardware. It is whether the whole procurement ecosystem is being redesigned to reduce dependence on U.S. technology.
For investors, the key point is that the story is not a single-quarter demand wobble. It is a structural repositioning of capital spending. If Chinese enterprises keep moving budget share toward local suppliers, Nvidia’s China exposure will likely remain constrained even if individual product approvals improve or shipment windows reopen. The competition is becoming less about one chip and more about which ecosystem can support large-scale AI deployment at acceptable cost.
Market Reaction and What the Survey Measures
The survey itself is the most concrete evidence in the story. Executives said domestic products will absorb 46% of their AI-accelerator spending over the next year, compared with 30% currently. That gap suggests a meaningful reallocation of demand away from imported chips and toward local vendors. It does not mean Nvidia’s hardware disappears from China overnight. It does mean that the budgetary center of gravity is moving in a direction that favors local suppliers.
The 80% over-budget reading is equally important. AI infrastructure spending can run above plan for many reasons: higher-than-expected chip prices, networking and power costs, data-center construction delays, software integration, or the need to buy more equipment to compensate for less efficient chips. But the common thread is that the AI buildout is expensive enough to force procurement teams to reconsider what they buy and from whom they buy it. In that environment, local suppliers can gain traction even when their products are not as advanced, because availability and support start to matter almost as much as benchmark performance.
That is one reason the China AI market has become more segmented. Frontier model training may still depend on the highest-end accelerators wherever they can be acquired, but inference, internal enterprise applications, and large-scale government or state-linked deployments can often be built on lower-tier or locally adapted silicon. Once buyers standardize around those systems, the switching costs move in the domestic suppliers’ favor. Nvidia may still be present, but it is no longer the default answer.
Market participants should also read the survey in the context of price sensitivity. Chinese buyers have been dealing with rising total system costs and uncertainty around import channels. If a domestic supplier can deliver enough performance for a meaningful discount, the procurement decision is easier. That is especially true in a market where policy goals can outweigh pure technical optimization. In other words, the budget share shift is not only a technology choice; it is a political-economy choice.
Why Nvidia’s China Position Is Eroding
The central reason is that U.S. export controls have pushed Chinese customers to build around domestic alternatives rather than waiting for access to the best Nvidia chips. Nvidia has already had to tailor products for China by stripping down performance in order to stay within the rules, but those adjusted chips have faced declining shipments and intensifying competition. The more restricted the market becomes, the more attractive domestic ecosystems become as long-term planning tools.
That matters because AI infrastructure procurement is sticky. Once a company or a state-backed data-center operator invests in a chip architecture, software stack, and supply chain, it tends to stick with that framework unless a material performance or cost gap opens up. Local suppliers therefore do not need to win every benchmark. They need to be good enough on performance and materially better on access, support, and policy alignment.
Huawei is the clearest beneficiary of that shift. The company has been making inroads in AI computing systems and has positioned its Ascend chips as a domestic alternative in a market where supply-chain independence carries strategic value. Other local chipmakers, software tool providers, and system integrators can benefit as well, but Huawei is the name most closely associated with China’s attempt to build an AI stack that can function with fewer foreign dependencies.
“We don’t expect an abrupt switch toward (Huawei’s) Ascend.”
That warning from Morningstar analyst Phelix Lee captures the nuance that makes the story more interesting than a simple replacement narrative. The transition away from Nvidia may be real, but it is unlikely to be linear. High-end AI training still favors top-tier hardware, and Chinese firms will continue to use whatever mix of foreign and domestic chips best fits the task. The key point is that the marginal buyer is increasingly looking local first.
The survey aligns with that pattern. It does not claim that Chinese demand for Nvidia has collapsed. It shows that the budget mix is moving toward domestic products. That is enough to alter long-term assumptions about China revenue contribution, product mix, and pricing power. Even modest changes in procurement share can have outsized consequences when they compound across hundreds of data centers and enterprise deployments.
There is also a second-order effect. As domestic suppliers gain more orders, they gain more field data, more engineering feedback, and more volume to improve manufacturing, software compatibility, and deployment practices. That learning loop can gradually close the gap further. The result is a feedback cycle: policy encourages local buying, local buying improves local products, and better local products make local buying easier.
How China’s AI Buildout Is Changing the Competitive Equation
The scale of China’s planned AI and data-center investment makes the current shift more consequential than a simple vendor swap. A 2 trillion yuan, five-year buildout is large enough to influence hardware standards, software ecosystems, and procurement norms across the country. When the government and state-linked companies are involved, buying decisions can become a tool of industrial policy, not just a procurement exercise.
That is why the 80% domestic-technology target in the broader plan matters. It signals that the goal is not merely to support a few local champions. It is to build a domestic supply chain capable of sustaining the whole AI stack, from chips and servers to power systems and interconnects. Once that architecture is in place, foreign suppliers face a tougher competitive environment even if they retain product advantages.
Nvidia has not disappeared from the story. The company still has a global lead in AI accelerators and still sells into markets where buyers prize performance above all else. But China is a special case because policy, geopolitics, and industrial strategy all point in the same direction. If a buyer in Shanghai or Shenzhen can use domestic hardware with less regulatory risk, less import uncertainty, and stronger local support, the incentive to remain dependent on Nvidia weakens.
That does not mean local chips are fully replacing Nvidia across the board. It means the market is splitting by use case. Frontier model training and the most demanding workloads may continue to lean on the best available imported chips where possible, while broader deployment shifts toward domestic hardware. For vendors, that creates a more complicated market: fewer easy wins, more price competition, and a stronger need to prove software compatibility and system-level reliability.
For policymakers, the result is more strategic autonomy. For domestic suppliers, it is a chance to scale. For Nvidia, it is a reminder that revenue from China may remain structurally capped even when demand for AI hardware elsewhere stays strong. The market is not just being fenced off; it is being redesigned.
“China’s strategy of pursuing technological self-sufficiency — and eventually exporting its technologies — is unlikely to change regardless of whether Nvidia can sell its chips in China.”
That view, from Counterpoint Research analyst Brady Wang, gets to the heart of the matter. The immediate issue is market share. The larger issue is whether China is using export restrictions as a forcing function to create a parallel AI infrastructure ecosystem. If that effort succeeds even partially, the competitive benchmark for global chipmakers changes from access to China to relevance inside a much more self-contained Chinese stack.
What To Watch Next
The next phase of the story will likely turn on three things. First, whether Chinese companies continue to increase the domestic share of AI-accelerator budgets in coming surveys and procurement data. Second, whether local suppliers can improve performance and software support fast enough to make domestic buying even more practical. Third, whether policy guidance from Beijing and state-linked buyers becomes more explicit about preferred chip stacks for large projects.
The other key question is how Nvidia responds. Product tailoring for China has already shown the limits of a purely export-led strategy when regulation intervenes. If demand keeps shifting toward local suppliers, Nvidia will have to balance the value of China exposure against the risk of undercutting its own product segmentation and global supply priorities. The company’s China business may remain important, but it is increasingly likely to be an option market rather than a core growth engine.
For the broader AI sector, the message is clear: hardware leadership is still valuable, but access, policy alignment, and ecosystem control are becoming just as important. In China, that combination increasingly favors local suppliers. The survey suggests the change is already underway, and the buildout plans show why it may last.
The market has spent years debating whether China can match U.S. AI chips on raw performance. The more immediate answer may be that it does not have to. If domestic hardware is good enough, available, and politically preferred, it can still win the budget.
Explore more exclusive insights at nextfin.ai.
