Nvidia’s China Share Falls to 34%, Huawei Hits 28% [2026]

Nvidia’s grip on China’s data-center AI-chip market slipped to 34% in the first half of 2026, down from a dominant position just two years ago, according to an IDC report reviewed by Reuters. Huawei, the Chinese tech giant that U.S. export controls have effectively pushed into Nvidia’s old backyard, now holds 28% of that same market after shipping 572,000 AI accelerators between January and June, up from more than 300,000 a year earlier. Nvidia’s own shipments fell to 678,000 units in the same window, a 44% drop year over year.

The numbers, first reported October 13, 2026, mark the clearest data point yet in a story that has been building for months: Washington’s export restrictions on advanced chips bound for China created a vacuum, and Huawei is filling it faster than almost anyone outside the company predicted. For a market that once generated billions in annual revenue for Nvidia, the shift carries real consequences for chip pricing, AI infrastructure build-outs, and the broader contest between U.S. and Chinese AI platforms.

·

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

The Numbers Behind Nvidia’s China Slide

The IDC figures give the clearest snapshot so far of how quickly the balance of power has moved. Nvidia shipped 678,000 AI accelerators into China during the first six months of 2026, a 44% year-over-year decline that reflects both tightened U.S. export licensing and a Chinese market increasingly willing to buy domestic. That volume translated into a 34% share of China’s data-center AI-chip market, according to the IDC data cited by Reuters.

Huawei’s trajectory runs in the opposite direction. Its 572,000 units shipped in the same six-month window gave it 28% of China’s overall AI-accelerator market, up from a base of just over 300,000 units a year earlier, per IDC. That is not a marginal gain. It is a near-doubling of shipment volume in twelve months, at a time when the overall addressable market in China was also expanding.

Put plainly, the gap between the two companies in China has narrowed from a wide Nvidia lead to a six-point spread. That is still a lead, but it’s a fraction of what it once was, and the trend line points toward further convergence rather than a snapback in Nvidia’s favor.

Metric (H1 2026)	Nvidia	Huawei
Units shipped in China	678,000	572,000
Year-over-year change	-44%	Up from 300,000+ (H1 2025)
Share of China AI-accelerator market	34%	28%
Primary driver	U.S. export licensing limits	Domestic substitution, assured supply
Source	IDC, via Reuters	IDC, via Reuters

Why US Export Controls Reshaped the Market

None of this happened in a vacuum. U.S. export controls restricted the supply of Nvidia’s most advanced data-center chips to Chinese buyers, a policy that has tightened and loosened in phases over the past two years as Washington balances national-security concerns against the risk of pushing China toward full chip self-sufficiency. Reuters’ reporting on the new IDC figures directly links Nvidia’s shrinking footprint to those restrictions, noting that supply limits on Nvidia products coincided with domestic rivals, led by Huawei, gaining ground.

The irony, several industry watchers have pointed out, is that export controls designed to slow China’s AI buildout may have accelerated the very outcome U.S. policymakers hoped to avoid: a fully domestic Chinese AI hardware stack that no longer depends on American silicon at all. Nvidia CEO Jensen Huang has made versions of this argument publicly. He has described China’s AI ecosystem as not an open question of whether it will develop, but a question of which country’s chip architecture it runs on.

That tension plays out against a backdrop of broader supply-chain realignment. Export-control pressure on chips has rippled into adjacent sectors too, including the EU’s own sweeping sanctions package targeting chips and drone components bound for Russia, part of a wider pattern of governments using semiconductor access as geopolitical leverage.

Huawei’s Ascend Line Gains Real Ground

Huawei’s rotating chairman, Eric Xu, has been unusually blunt about the company’s ambitions in AI silicon. Speaking about the Chinese market, Xu said it’s pretty hard to collect data about the market share of Nvidia in China, but based on the data we have collected, Ascend has surpassed Nvidia, according to comments reported by The Register. Huawei did not provide data to support that estimate, and it should be treated as a company claim rather than an independently verified figure. The IDC numbers reviewed by Reuters, which put Huawei at 28% versus Nvidia’s 34% for the first half of 2026, do not support the stronger claim that Ascend has already overtaken Nvidia outright, though they do confirm the gap has narrowed substantially.

Xu’s broader pitch to Chinese buyers centers on reliability rather than raw performance. He argued that even though Huawei’s chips may be less advanced, at least their supply is assured, so buyers don’t have to worry about chip supply day in and day out, per The Register. That is a notable admission: Huawei is not claiming performance parity with Nvidia’s top-end data-center silicon. It is selling certainty, in a market where Nvidia’s availability has been hostage to shifting export rules.

Xu also framed Huawei’s strategy in explicitly national terms, saying that for the Chinese government, the domestic industry, and Huawei itself, the path forward is undoubtedly to push for full self-sufficiency in chips and the entire semiconductor value chain, according to the same Register report. That statement lines up with years of Chinese industrial policy aimed at reducing dependence on foreign chip suppliers, a policy goal that export controls have arguably sharpened rather than blunted.

The Non-GPU Chip Makers Rounding Out China’s Ecosystem

Huawei is not operating alone. According to the IDC data, Huawei, T-Head (Alibaba’s chip design unit), Kunlunxin (Baidu’s AI chip arm), and Cambricon together accounted for 95% of China’s non-GPU AI accelerator segment in the period measured. That figure matters because it shows the depth of China’s homegrown chip bench, not just Huawei’s individual rise. Even if Huawei stumbled, three other credible domestic suppliers are positioned to absorb demand.

This is a meaningfully different competitive structure than the one Nvidia faced even three years ago, when Chinese AI chip alternatives were widely viewed as years behind in both performance and software tooling. The build-out of China’s domestic AI data-center footprint, including the kind of facility expansion covered in reporting on Taiwan-based ProLogium’s pivot toward AI data-center supply contracts, has given these non-GPU chipmakers a growing customer base to sell into, independent of whether Nvidia chips are available at all.

IDC’s broader forecast for 2026 adds useful context: total AI-accelerator shipments in China were expected to grow 19% for the full year, while GPU shipments specifically were expected to fall 9%. Read together, those two numbers describe a market that is still growing overall but shifting composition, away from the GPU architecture Nvidia specializes in and toward a more diverse mix of non-GPU accelerator designs from domestic vendors.

Conflicting Market-Share Estimates Show How Murky the Picture Really Is

One thing that stands out across months of reporting on this topic is how much the specific percentages vary depending on which research firm, which time period, and which market definition is used. That is worth laying out plainly rather than smoothing over, because it shapes how much confidence anyone should place in any single number, including the headline 34% and 28% figures from the newest IDC report.

Source	Period Covered	Nvidia Share	Chinese Rivals Share	Reported By
IDC	H1 2026	34%	28% (Huawei alone)	Reuters, Oct. 2026
IDC	Full-year 2025	Not specified	Nearly 41% (combined Chinese GPU/AI chipmakers)	Reuters, April 2026
Reuters reporting	Full-year 2025	55% (2.2 million units shipped)	Not specified	Reuters, May 2026
Bernstein estimate	2025, with 2026 forecast	~40% in 2025, predicted to fall near 8% in 2026	~40% in 2025, predicted to grow further in 2026	LA Times, June 2026
Huawei (Eric Xu, self-reported)	Unspecified	Not specified	Claims Ascend “has surpassed Nvidia”	The Register, September 2026

The spread between a 55% Nvidia share in one Reuters report and a roughly 40% share in the Bernstein estimate, both describing 2025, illustrates the measurement problem. Different firms count different product categories, include or exclude AMD and other foreign vendors, and draw from different customer surveys. What is consistent across every version of the story, though, is the direction of travel: Nvidia’s share has been falling and Huawei’s has been rising, across every methodology anyone has published.

Notably, Bernstein’s June 2026 forecast that Nvidia’s share would shrink to around 8% by the end of the year looks far too pessimistic against the actual IDC figure of 34% for the first half. That gap is a useful reminder that even well-resourced equity research estimates have struggled to model how fast, or how far, this shift would go.

What Nvidia’s Leadership Is Saying

Nvidia CEO Jensen Huang has been candid in public remarks about the scale of the retreat. The demand in China is quite large, Huang said, describing Huawei as very, very strong, adding that Huawei will likely have an extraordinary year and that the local ecosystem of chip companies are doing quite well, because we’ve evacuated that market, in comments reported by CNBC. Huang’s phrase, that Nvidia has largely conceded that market to them, is a striking admission from a sitting CEO about ground lost to a direct competitor.

Huang has also framed the stakes in geopolitical terms rather than purely commercial ones. He has argued that the real question is not whether China will have AI, since it already does, but whether one of the world’s largest AI markets will run on American platforms, according to comments reported by 247wallst.com. That framing positions the China AI-chip fight as something closer to a standards war than a simple sales contest, where losing market share today could mean losing software-ecosystem lock-in for years afterward.

Nvidia has not disclosed a detailed public roadmap for winning back Chinese customers, and the company’s broader product strategy increasingly looks tilted toward markets outside China, including the enterprise AI infrastructure push reflected in Nvidia’s recent decision, covered separately, to discontinue the RTX 5090 and redirect its GB202 chip toward the RTX Pro lineup, a move that shifts manufacturing capacity toward higher-margin professional and enterprise buyers rather than the consumer or China-exposed segments.

Historical Context: How Nvidia Built Its China Dominance

It is worth remembering how recently Nvidia’s China position looked unassailable. As recently as 2025, Chinese GPU and AI chipmakers combined held under half the local market, and earlier Reuters reporting from April 2026 described Chinese rivals capturing nearly 41% of China’s AI accelerator server market for the full year 2025, itself framed at the time as a dramatic erosion of what had been Nvidia’s dominant position in one of its most important overseas markets.

Total AI accelerator card shipments in China reached roughly 4 million units in 2025 across Nvidia, AMD, and Chinese vendors combined, according to that same IDC-sourced reporting. That scale explains why the fight over market share matters so much financially. Even a swing of a few percentage points translates into hundreds of thousands of units and, at data-center chip price points, billions of dollars in revenue moving between competitors.

Nvidia’s historical strength in China rested heavily on its CUDA software ecosystem, which for years gave it a moat that Chinese rivals struggled to replicate regardless of raw silicon performance. The newer dynamic, where Huawei is explicitly selling supply certainty rather than claiming to match Nvidia’s software tooling, suggests that moat is eroding for reasons that have less to do with chip design and more to do with export policy making Nvidia an unreliable supplier in the eyes of Chinese buyers.

Market Impact: What This Means for Nvidia’s Business

China was once one of Nvidia’s largest data-center markets outside the United States, and a 44% year-over-year drop in shipment volume is not a figure that disappears quietly into a diversified revenue base. The company has leaned harder into other growth areas, including cloud AI infrastructure deals with hyperscalers, enterprise AI chips, and the kind of GPU pricing dynamics tracked in comparisons of AWS, Azure, and Google Cloud’s GPU pricing and capacity allocations, where demand for Nvidia silicon outside China remains extremely strong.

That diversification matters because it means the China shortfall, while significant, is not necessarily an existential threat to Nvidia’s overall growth story the way it would have been five years ago, when China represented a much larger share of total data-center chip revenue. Still, ceding structural ground in one of the world’s largest AI markets carries long-term costs beyond this year’s shipment numbers, particularly if Chinese AI developers standardize their software stacks around Huawei’s Ascend architecture rather than Nvidia’s CUDA.

The broader AI hardware and software pricing environment has also grown more competitive on multiple fronts simultaneously. The same pressures reshaping chip market share in China echo a pattern seen in AI software pricing as well, where releases like Claude Haiku 5.5’s steep price cut touched off a broader AI pricing war. Competitive pressure on Nvidia in China is arguably part of the same larger trend: AI infrastructure costs, whether measured in chips or in inference pricing, are under sustained downward pressure across the entire stack.

Competition: Huawei, T-Head, Kunlunxin, and Cambricon vs Nvidia

Nvidia’s China competition is no longer a single-company contest. Huawei leads among the non-GPU vendors, but it shares that 95% non-GPU segment share with three other domestic suppliers, each tied to a different corner of China’s tech industry. T-Head is Alibaba’s chip design arm, Kunlunxin belongs to Baidu, and Cambricon operates as an independent, publicly traded chip designer. Each brings a different customer base: Alibaba and Baidu can funnel their own massive cloud and AI businesses toward in-house silicon, while Cambricon competes more directly for third-party sales.

That multi-vendor depth is arguably more important to Nvidia’s long-term China position than Huawei’s individual share number. A single dominant domestic competitor could, in theory, be out-innovated or out-priced over time. Four credible domestic suppliers splitting 95% of a fast-growing non-GPU segment represents a much harder competitive environment to dislodge, because it reduces the single-point-of-failure risk that Chinese AI buyers might otherwise worry about when relying on one alternative supplier.

Globally, Nvidia continues to face a more traditional rival in AMD, whose memory and GPU pricing moves, including AMD’s recent GDDR6 price increase, contrast with Nvidia holding firm on GDDR7 pricing, as detailed in coverage of AMD’s price hike decision. That dynamic plays out mostly outside China, showing that Nvidia is now fighting a multi-front competitive battle: AMD in global GPU markets, and an entirely separate cluster of Chinese chipmakers inside China specifically.

The Global Ripple Effect Beyond China

What happens inside China’s AI chip market rarely stays contained to China. If Huawei, T-Head, Kunlunxin, and Cambricon continue gaining share domestically, their combined manufacturing scale and R&D budgets grow too, potentially positioning them to compete for AI infrastructure contracts in markets outside China over time, particularly in regions less constrained by U.S. export rules.

There’s also a software-ecosystem dimension that extends well past chip sales figures. As Chinese AI labs increasingly train and deploy models on Huawei’s Ascend hardware rather than Nvidia GPUs, the tooling, libraries, and optimization work built around Ascend compounds over time, making it progressively harder for Nvidia to win those customers back even if export rules eventually loosen. This mirrors concerns Huang himself has voiced publicly about losing not just a sale, but a platform.

The broader AI open-source movement adds another layer. As more AI model developers, including the kind of open-source momentum described in reporting on open-source AI bets from companies like Beam and Mistral, ship models that run efficiently across varied hardware architectures rather than being locked to a single vendor’s software stack, the competitive importance of any single chipmaker’s software moat, Nvidia’s included, could gradually diminish.

IDC’s 2026 Outlook for China’s AI Chip Market

Looking at the full year, IDC’s forecast paints a market that keeps growing even as its composition shifts. Total AI-accelerator shipments in China were expected to grow 19% in 2026, a healthy expansion rate for any hardware category. But that growth was not expected to be evenly distributed: GPU shipments specifically, the category Nvidia dominates, were forecast to fall 9% even as the overall market expanded.

That divergence, overall market growing while the GPU sub-segment shrinks, is the clearest statistical expression of the structural shift underway. It implies that nearly all of the incremental growth in China’s AI chip market in 2026 is being captured by non-GPU architectures from domestic vendors, rather than by Nvidia’s traditional GPU designs, even as Chinese AI compute demand keeps climbing overall.

Predictions: Where the Nvidia-Huawei Rivalry Goes From Here

Based on the trajectory visible in this and prior IDC-sourced reports, several outcomes look increasingly likely over the next several quarters.

• Huawei’s share will likely keep climbing in the second half of 2026, given the shipment momentum already visible and the company’s stated push toward chip self-sufficiency.
• Nvidia’s China share is unlikely to collapse to the single digits that Bernstein’s June 2026 forecast suggested, since the actual H1 2026 figure of 34% already far exceeds that prediction, but further gradual decline toward Huawei’s level, or below it, remains plausible if export rules stay tight.
• The non-GPU segment led by Huawei, T-Head, Kunlunxin, and Cambricon will likely keep expanding faster than the GPU segment, continuing the pattern IDC forecast for full-year 2026.
• Expect continued public sparring over whose market-share figures are accurate, with Huawei likely to keep making stronger claims than third-party researchers can independently verify.
• Nvidia will likely continue pivoting its product roadmap, including moves like the RTX 5090 discontinuation and GB202 reallocation toward enterprise-grade RTX Pro silicon, toward markets less exposed to Chinese export restrictions.

What This Means for the Broader AI Hardware Industry

For an industry that has spent the past three years treating Nvidia’s dominance as close to inevitable, the China numbers are a reminder that dominance built on export-dependent supply chains carries its own fragility. Export controls can protect national-security interests, but they can also, as this case shows, hand a motivated domestic competitor exactly the opening it needs to scale manufacturing, refine its software stack, and win customer trust on reliability grounds alone.

For buyers and AI infrastructure planners outside China watching this play out, the lesson is less about China specifically and more about supply-chain concentration risk generally. Any company anywhere that depends heavily on a single chip architecture, whether Nvidia’s or anyone else’s, is watching a live case study in how quickly that dependence can become a liability when politics intervenes in a market that used to run on pure performance and price.

Frequently Asked Questions

What is Nvidia’s current market share in China’s AI chip market?
Nvidia held 34% of China’s data-center AI-chip market in the first half of 2026, according to an IDC report reviewed by Reuters, down sharply from its earlier dominant position.

What is Huawei’s market share in China’s AI chip market?
Huawei held 28% of China’s overall AI-accelerator market in the first half of 2026, per the same IDC data, after shipping 572,000 units compared with more than 300,000 a year earlier.

Has Huawei’s Ascend chip actually overtaken Nvidia in China?
That claim has been made by Huawei’s rotating chairman Eric Xu, who said Ascend has surpassed Nvidia based on Huawei’s own data collection. Huawei did not release supporting data, and the independently verified IDC figures show Huawei at 28% versus Nvidia’s 34% for H1 2026, which does not confirm Xu’s stronger claim.

Why is Nvidia losing market share in China?
U.S. export controls have restricted the supply of Nvidia’s most advanced AI chips to Chinese customers, creating supply uncertainty that domestic rivals like Huawei have exploited by offering assured, if less advanced, chip supply.

Who else besides Huawei is competing for China’s AI chip market?
T-Head (Alibaba), Kunlunxin (Baidu), and Cambricon round out the group that, together with Huawei, accounted for 95% of China’s non-GPU AI accelerator segment, according to IDC.

Is China’s overall AI chip market still growing?
Yes. IDC expected total AI-accelerator shipments in China to grow 19% in 2026, even though GPU shipments specifically, Nvidia’s core category, were expected to fall 9% over the same period.

How does this affect Nvidia’s overall business outside China?
China’s declining share is significant but not necessarily existential for Nvidia, since the company has diversified into other data-center and enterprise AI markets where demand for its chips remains strong, including through product shifts like the RTX 5090 discontinuation in favor of enterprise-focused RTX Pro silicon.

Why do different reports give different Nvidia market-share numbers for China?
Research firms use different methodologies, time periods, and product-category definitions. IDC’s own figures show Nvidia at 34% for H1 2026, while separate reports have cited figures ranging from roughly 40% to 55% for full-year 2025, reflecting genuine measurement differences rather than a single settled number.

Related Coverage

• Yandex Outage: Bot Traffic Hits 0.4%, Not Proof [2026]
• Anthropic Cuts AI Distillation Attacks by 99% [2026]
• Meta Muse Wonder AI Hits Frontier in 14 Months [2026]
• Smartphone Shipments Drop 20%, Worst Q3 in 14 Years [2026]
• Honda Super-ONE EV Hits $36,900 Amid Charging Buzz [2026]