Nvidia's AI chips collapse in China as Huawei takes over

What the raw numbers reveal

Barely three years ago, Nvidia held roughly 95% of China's artificial intelligence chip market, according to founder Jensen Huang's own words as reported by the Associated Press and picked up by the Los Angeles Times. Today, that figure has all but vanished in some segments, as Chinese manufacturer Huawei captures a growing share of the domestic market. Analysis firm Bernstein puts Nvidia's China market share at around 40% in 2025, with a projected drop to just 8% this year.

This reversal is no accident. It stems directly from the export controls imposed by Washington on advanced semiconductors, a national security policy that pushed Beijing to accelerate its pursuit of technological self-sufficiency. Sorting out what is proven from what remains speculation is the point of this fact-check.

Why this story deserves rigorous verification

The narratives circulating about Chinese AI chips often mix solid facts with smuggling rumors, conflicting market-share estimates, and criminal accusations still under investigation. One striking example: Reuters' revelations that Huawei allegedly stockpiled 2 to 3 million chips manufactured by TSMC, obtained through an alleged diversion scheme involving Chinese company Sophgo, already on the American blacklist.

This fact-check separates what is confirmed by primary sources — official statements, analyst reports, court documents — from what remains an unproven allegation. I won't lie to you: in this story, the line between verified fact and plausible theory is sometimes thin, and I'd rather admit that than make the call for you without solid proof.

What the primary source says

Jensen Huang confirmed this figure himself in an interview with the Associated Press, picked up by the Los Angeles Times on June 29, 2026: "We were in China for 30 years, and before the export control banned Nvidia from China, we had about 95% market share, so we were in full competition." This statement comes directly from the person at the center of it, which makes it a top-tier source, though it also serves his interests in the debate over easing American restrictions.

Verdict: TRUE, with the caveat that this figure represents a historical average across three decades of commercial presence, not a frozen snapshot in time. The subsequent collapse is likewise widely corroborated by several independent analyst firms.

What follows: is the drop as steep as claimed

According to Bernstein, Nvidia's share fell to around 40% in 2025, and is expected to drop to 8% in 2026. These figures, cited by the Los Angeles Times, apply specifically to the advanced AI chip segment, not the entire Chinese semiconductor market, where Nvidia and AMD still maintain a presence in certain less sensitive segments.

Other data, notably from the firm IDC for the full year 2025, puts Nvidia's share around 55% of a total market of roughly 4 million cards, with domestic Chinese suppliers — led overwhelmingly by Huawei — capturing between 41% and 49%. These discrepancies between firms are not a sign of manipulation: they simply reflect different methodologies and calculation scopes, a classic problem when trying to measure a market as opaque as China's semiconductor sector.

Huawei's production numbers

According to data compiled by IDC, Huawei reportedly shipped around 812,000 AI accelerators in 2025, close to 49% of Chinese-brand cards and around 20% of the total market — more than every other Chinese supplier combined. The Los Angeles Times confirms that Huawei "now leads domestically" according to analyst Antonia Hmaidi, who states plainly that "Nvidia has definitely lost significant ground to Huawei."

The flagship product behind this breakthrough is the Ascend 950 series, seen by several industry analysts as comparable, in certain respects, to Nvidia's H200 — one of the most powerful chips in the American lineup. Analyst Phelix Lee tempers that view, however: "We don't expect an abrupt shift toward Ascend," noting that demand still outstrips available supply in China, according to analyst Rui Ma.

Verdict on this claim

The verdict is largely true with caveats. Huawei has indeed overtaken or matched Nvidia in several segments of the domestic Chinese market, but the transition is neither complete nor instantaneous. Major Chinese tech companies and universities still want chips like the H200, partly for research purposes, which fuels a parallel market for circumventing the restrictions.

He Tingbo, a technical leader at Huawei, sums up the moment's ambivalence: "We've found pretty good solutions." But she also admits to lingering uncertainty: "Who can move faster? Huawei or other companies? I don't know the answer. I think only time will tell." That honesty from a Huawei executive deserves to be highlighted: even the architects of this Chinese rise acknowledge the technological uncertainty that remains, and I find it refreshing that she doesn't give in to easy propaganda.

What the investigations reveal

This claim, featured at the top of this story, comes from a broader investigation into attempts to circumvent American restrictions. The U.S. Commerce Department has opened an investigation into TSMC, the Taiwanese semiconductor manufacturing giant, to determine whether chips destined for Chinese company Sophgo — itself on the blacklist — ended up in Huawei's Ascend 910B AI processor. According to information relayed by Reuters, TSMC could face a fine exceeding a billion dollars if the accusations are confirmed.

TSMC categorically denies any direct shipments to Huawei since September 2020, when American restrictions first took effect against the Chinese company. Because the semiconductor supply chain is global and complex, tracing the exact origin of each chip back to its end user is painstaking investigative work, often spanning several years.

What this fact-check can and cannot confirm

The verdict here is partially verifiable, investigation ongoing. The existence of a Commerce Department investigation is confirmed by reliable journalistic sources. The precise figure of "2 to 3 million chips" cited in some press summaries has not been independently confirmed by an official primary source at the time of writing, and I refuse to present it as an established fact until a court document or government statement formally corroborates it.

This is exactly the kind of nuance I need to own as a columnist: reporting that an accusation exists and circulates in the trade press is one thing, claiming it's proven is another. Reasonable doubt protects the credibility of this entire story, and I'd rather be cautious than sensational.

The losses in numbers

Nvidia recorded an inventory and development loss of $5.5 billion directly tied to the export restrictions, a figure confirmed by several financial analyses, notably reported by Yahoo Finance. The American giant had designed its H20 chips, a version with reduced computing power, specifically to keep selling in China without violating American restrictions. Those shipments were already declining steadily last year.

As for the H200, its situation remains ambiguous: at a recent shareholder meeting, Jensen Huang admitted the product has "generated no revenue yet, and we don't know if imports will be allowed into the country." Beijing maintains strategic ambiguity on this issue, which complicates any reliable forecast for Nvidia.

The scale of the shock to China's tech ecosystem

China's cyberspace administration reportedly ordered domestic giants like ByteDance and Alibaba to completely halt their Nvidia chip purchases, a measure reported in September 2025 that shows Beijing's determination to accelerate self-sufficiency at any cost. Researcher Paul Triolo even points to a "significant collaborative effort between DeepSeek and Huawei" to train future AI models on domestic hardware — DeepSeek has confirmed that its V4 model, launched in April, was adapted for Huawei's advanced Ascend chips.

The overall verdict on this dynamic: largely confirmed. Nvidia's financial losses and China's accelerating self-sufficiency are two sides of the same coin, documented by consistent financial and journalistic sources.

What the reality of the supply chain reveals

This is where the Chinese narrative of "total technological victory" falls apart under scrutiny. The advanced semiconductor supply chain remains deeply global, and no country today can manufacture a cutting-edge AI chip entirely on its own. Nvidia itself depends on the extreme ultraviolet lithography machines made by Dutch company ASML, which in turn rely on American components and technology. TSMC uses these machines to produce most of Nvidia's advanced chips in its Taiwanese factories.

China remains formally barred from buying Nvidia's most powerful AI chips and ASML's EUV lithography machines. This structural dependency means that even Huawei's rapid gains partly rely on older-generation Chinese semiconductors, technologically less advanced than their most recent Western equivalents.

What the West should take away from this story

Analyst Wang, cited by Counterpoint, sums up the real trajectory: "China's strategy of pursuing technological self-sufficiency — and eventually exporting its technology — likely won't change, whether or not Nvidia can sell its chips in China." That's a clear warning: even if Washington tightened the screws further, Beijing would keep investing massively in its technological independence, with the eventual ambition of exporting these solutions to other markets, particularly in Southeast Asia.

Here is where I state my judgment as a columnist plainly: the West cannot afford to win a communications battle while losing the industrial war. Every year Huawei closes the gap is a year in which America's strategic advantage in artificial intelligence shrinks, and pretending otherwise would be denial.

What recent smuggling cases show

Several recent court cases in the United States have exposed organized networks smuggling Nvidia chips into China, circumventing export restrictions through third countries in Southeast Asia. The Los Angeles Times notes that "several recent cases tied to smuggling Nvidia's AI chips into China show the appetite for its technology," an appetite that directly contradicts the narrative of a fully self-sufficient China indifferent to American products.

This reality creates an interesting tension in the debate: if China no longer needed American chips, why would entire networks risk criminal prosecution to keep smuggling them in illegally? The answer comes down to one word: performance. Huawei's best chips, however impressive, still don't match Nvidia's most advanced products for certain large-model training workloads.

The verdict on this claim

The verdict is true and documented. The court cases and customs seizures reported by several American outlets confirm the existence of an active parallel market. This does not contradict Huawei's rise, but it does nuance the idea of a total, definitive shift toward domestic Chinese solutions.

The stakes beyond semiconductor trade

Beyond the sales figures, this technological rivalry fits into a broader geopolitical contest between the United States and China, alongside other strategic rivals of the West like Russia, Iran, and North Korea. The ability to produce and deploy cutting-edge AI chips will determine who dominates future autonomous military systems, mass surveillance capabilities, and major scientific breakthroughs in the decades ahead.

The global AI chip market is expected to reach roughly $67 billion by 2030, according to projections cited by several specialized analysts, which explains the intensity of the financial and strategic stakes behind every percentage point of market share won or lost by Nvidia and Huawei.

What the West must do to avoid losing this race

Western governments must choose between two approaches: maintaining strict restrictions that paradoxically accelerate Chinese self-sufficiency, or easing certain exports to retain commercial influence over the Chinese tech ecosystem while curbing its most dangerous applications. Neither option is perfect, and decision-makers in Washington will need to choose with a long-term vision rather than according to electoral cycles.

I'll say it plainly: China doesn't need to win this battle quickly, it just needs to never give up the race. That's where the West, with its four-year political cycle, must prove it can sustain a coherent industrial and strategic effort over an entire decade, not just the length of a presidential term.

What's confirmed, what remains uncertain

At the end of this verification, several central elements of the narrative are confirmed by multiple credible sources: Nvidia's dramatic loss of market share in China, Huawei's rapid rise with its Ascend series, and the concrete financial losses suffered by the American giant. Other elements, notably the exact scale of the alleged diversion scheme via Sophgo and TSMC, remain under investigation and deserve to be followed cautiously in the coming weeks.

This story illustrates a broader reality: the technological rivalry between Washington and Beijing will not be settled in a single quarter. It plays out over years, through restrictions, workarounds, forced innovation, and financial losses shared by both sides.

Why this battle goes beyond a simple commercial issue

Supremacy in artificial intelligence isn't just a matter of profits for Nvidia or national pride for China. It's a matter of security, future military power, and global technological leadership for decades to come. The West must choose between a coherent long-term restriction policy and the temptation to give in short-term to pressure from financial markets demanding renewed access to the Chinese market for American companies.

This fact-check is not meant to settle that political debate, but to give readers the verified facts they need to form an informed opinion, without giving in to either American triumphalism or Chinese propaganda.

By Maxime Marquette, columnist

Editorial stance

I am not a journalist but a columnist and analyst. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies.

I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. I claim analytical clarity, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events.

Method and sources

This text keeps the basic distinction between verified facts and interpretive analysis. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under Sources, never here.

Primary source families used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies.

Secondary source families: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations.

When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under Sources.

Nature of the analysis

The analyses, interpretations and perspectives in the analytical sections of this article are a critical, contextual synthesis based on the information available, the trends observed and the expert comments cited in the sources consulted.

My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text.