Nvidia loses entire China market as Huawei AI chip sales surge 60% to $12 billion

Jensen says Nvidia now has 'zero percent' market share in China -- says US export policy 'has already largely backfired'

Nvidia CEO Jensen Huang said that the company's market share of AI accelerators in China has now dropped to 0%. The drop is staggering, given that the company owned a lion's share of China's AI accelerator market just about two years ago. "In China, we have now dropped to zero," said Jensen Huang in an interview with the Special Competitive Studies Project, a bipartisan initiative by American lawmakers aimed at ensuring long-term competitiveness of the U.S. "Conceding an entire market the size of China probably does not make a lot of strategic sense, so I think that has already largely backfired. Maybe it made sense at the time, but I think the policy really needs to be dynamic and needs to stay with the times. I think it would be fairly safe to say that having American chip companies and other companies in China makes a lot of sense." Earlier this year, Bernstein estimated that Nvidia's share of China's AI GPU market could fall from 66% in 2024 to roughly 8% in the coming years, both due to restrictions imposed by the U.S. government and because domestic vendors are moving to cover up to 80% of demand. According to Huang, this happened much later, though, again, he only talks about Nvidia's direct sales to Chinese customers. Meanwhile, Huang warns that even without leading AI GPUs and software stacks developed in America, China remains a formidable competitor when it comes to frontier AI models. "American companies win around the world," Huang said. "The argument there is that across the five-layer cake, there's one particular layer that is too important because in the others, China can get ahead. They have cheaper energy. They have incredible talent. So, they [have] the number of science and math experts, and as a result of that, the number of AI researchers in China is quite extraordinary, it's one of their national treasures." Given the situation, Huang also contends that U.S. export controls may be strategically counterproductive. He argues that conceding a market of that scale accelerates China's push toward self-sufficiency, while continued participation of American companies in that market would help extend the global reach of the American AI technology stack. Indeed, Chinese developers are increasingly relying on local hardware, with companies like Huawei, Cambricon, Moore Threads, and MetaX advancing both silicon and software. In software -- the so-called CUDA moat in particular -- remains the main frontier of American AI technology in China that local companies have yet to conquer. Ultimately, Huang warns that fear-driven narratives and export controls could slow the deployment of AI more broadly as China and other regions embrace it more aggressively as an economic tool. Long-term leadership will depend less on restricting global rivals and more on ensuring that the American AI ecosystem dominates globally, according to Huang. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Huawei's AI chip sales surge as Nvidia stalls in China

Huawei is set to capture the largest share of China's AI chip market this year, with sales jumping by at least 60 per cent amid strong demand from Chinese companies seeking domestic alternatives to Nvidia. China's tech groups have placed large orders for Huawei's latest Ascend processor 950PR, according to two people with knowledge of the matter, as the Shenzhen-based company races to make inroads into a sector long dominated by the $5.1tn US chip group. Huawei expects revenues from its AI chips to reach about $12bn this year based on orders it has already received, up from $7.5bn in 2025. The majority of orders this year are for 950PR, which entered mass production in March. It also plans to launch an upgraded version of 950DT in the fourth quarter, the people said. The Chinese tech group has aggressively expanded its chipmaking capabilities as Nvidia continues to face regulatory barriers in Washington and Beijing. While Nvidia chief Jensen Huang said in March the company had finally received US licences to sell its H200 chips to China, no shipments have yet been made due to regulatory issues, according to two people with knowledge of the matter. Beijing has told Chinese tech companies to support domestic producers and limit their use of Nvidia chips to overseas, one of the people said. Meanwhile US regulators require that all Nvidia chips ordered by Chinese clients only be used in China, according to the person. This has led to no customs clearance for H200 in China, which is required for shipments. Huawei's AI chip revenue forecast could rise if its production ramp-up exceeds projections, the people said. The Chinese tech group manufactures most of its AI chips with Semiconductor Manufacturing International Corporation (SMIC), China's leading fab. Later this year two additional fabrication plants dedicated to its needs are scheduled to launch. While Huawei's latest chips still lag Nvidia's most advanced products by at least two generations, the Chinese company has steadily improved the performance and efficiency of its AI chips. The Chinese tech group has positioned its latest 950PR processors as the hardware of choice for domestic companies running "inference", the computation LLMs use to generate responses. Huawei is betting that inference, which is technically less challenging than model training, will be a bigger future source of demand as AI applications such as agents become more widespread. The tech group is also leveraging its networking capability to build powerful computing clusters, offsetting weaker single-chip performance. Earlier this month DeepSeek said it worked with Huawei's 950PR for its latest V4 model, which features improved inference efficiency at reduced costs. But the majority of V4's training was still done on Nvidia's chips, according to people familiar with the matter. "The day that DeepSeek comes out on Huawei first, that is a horrible outcome for our nation," Nvidia's Huang said in a recent interview with podcaster Dwarkesh Patel. It could lead to a scenario where "AI models around the world are developed and they run best on non-American hardware", he added. A key challenge for Nvidia's Chinese competitors has been building a software system to rival the US chipmaker's Cuda, which remains dominant globally. Huawei's Cann, started a decade after Cuda, has drawn complaints for being difficult to use, which greatly increases clients' operating costs. While improvements are being made, its disadvantage compared to Nvidia's Cuda remains significant, developers say. Morgan Stanley forecasts China's AI chip market will reach $67bn in 2030, with 86 per cent expected to be supplied by Chinese players. It estimates the market to be about $21bn this year from domestic suppliers. China's AI chip sector was previously dominated by Nvidia, which sold $17.1bn in fiscal year 2025, mostly from H20, a watered-down product tailored for China to meet US export controls. Since the US banned the sales of the H20 to China a year ago, Nvidia has struggled to re-enter the market with another AI chip that would satisfy both US and Chinese regulators' concerns. "Two structural forces shape the market: a rapid rise in AI inference demand . . . and persistent export controls, making localization a long-duration feature of China's AI compute market rather than a temporary policy response," wrote Morgan Stanley analysts. Huawei did not respond to a request for comment. Nvidia declined to comment. With additional reporting from Cheng Leng in Beijing and Demetri Sevastopulo in Washington

Exclusive: Big Chinese tech firms scramble to secure Huawei AI chips after DeepSeek V4 launch, sources say

BEIJING, April 29 (Reuters) - Demand for Huawei's Ascend 950 AI chips has surged following the release of DeepSeek's V4 artificial intelligence model that runs on the Shenzhen-based tech firm's chips, with major Chinese internet firms rushing to secure orders, three people familiar with the matter said. China's biggest internet firms including ByteDance, Tencent (0700.HK), opens new tab and Alibaba (9988.HK), opens new tab are reaching out to Huawei about new chip orders, said the sources, who are familiar with the procurement discussions. Companies specializing in cloud computing and graphics processing unit (GPU) rental services are also scrambling to place orders, two of the sources added, without providing the names of the firms. While the 950PR significantly outperforms Nvidia's H20 chip - the most powerful chip Nvidia was permitted to sell in China until Beijing blocked its import last year - it still trails the American firm's H200, a more advanced processor that has been caught up in regulatory limbo. Despite U.S. and Chinese approvals for exports, the H200 has yet to be shipped to China as Beijing and Washington remain at odds over the conditions governing its sale, providing an opportunity for Huawei to sell its ⁠semiconductors. The 950PR represents a breakthrough for Huawei after years of struggling to win large orders from China's tech sector. Customer testing of the chip went well earlier this year, with firms including ByteDance and Alibaba planning orders after samples were distributed in January, Reuters reported in March. Huawei, ByteDance, Alibaba and Tencent did not respond to Reuters requests for comments. DEEPSEEK FRENZY The scramble for Huawei's chips underscores how DeepSeek's V4 release last week has turbocharged demand for domestic Chinese AI hardware as U.S. export controls continue to restrict access to Nvidia's most advanced processors. It is also an endorsement of the performance of Huawei's chips so far. DeepSeek's decision to optimize its V4 specifically for Huawei's chips marks a strategic shift away from American semiconductor dependence and more towards China's homegrown AI gear, which is a priority for Beijing as it seeks tech supremacy. Last week, Huawei said its Ascend supernode infrastructure, built on the Ascend 950 series chips, would fully support DeepSeek V4 models and that the entire Ascend SuperNode product line had been adapted for V4 inference, referring to the process of using a trained AI model to answer queries and execute tasks. Among Chinese chipmakers, ⁠Huawei's Ascend 950 series - specifically the 950PR variant - is the only domestic chip to support a technique that processes AI calculations in a more compressed numerical format, allowing it to handle more computations per second at a lower cost. Highlighting the rush for access, Alibaba Cloud's Bailian platform made DeepSeek V4 available on the day that it was released, offering both the V4-Pro and V4-Flash variants with pricing matching DeepSeek's official rates. Tencent Cloud launched V4 preview services on its TokenHub platform on the same day, deploying the model on both domestic nodes and its Singapore international gateway to serve global users. The rapid deployment by major ⁠cloud platforms means millions of users and developers can now access V4, sharply increasing the volume of AI queries that need to be processed - and with it, demand for the underlying chips. SUPPLY CONSTRAINTS PERSIST DeepSeek, which is offering developers a 75% discount on its new model until May 5, said V4-Pro pricing could decline significantly in the second half of 2026 once Huawei's Ascend 950 supernodes "ship at scale." However, the company acknowledged that ⁠constraints would persist until production ramps up, reflecting the tight supply of high-end homemade AI chips. DeepSeek's V4 includes two versions: V4-Pro with 1.6 trillion total parameters and V4-Flash with 284 billion parameters, both supporting a one-million-token context window. The models are available as open-source releases under the permissive MIT open-source licence, which allows companies to freely use, modify and commercialise the models. However, ⁠output of the 950 is expected to fall short of demand due to U.S. export restrictions on advanced chipmaking tools that prevent China from acquiring cutting-edge manufacturing equipment. Huawei planned to ship around 750,000 units of the 950PR this year, with mass production beginning in April and full-scale shipments starting in the second half of 2026, according to people familiar with the plans. Reporting by Beijing and Hong Kong newsrooms; Editing by Anne Marie Roantree and Thomas Derpinghaus Our Standards: The Thomson Reuters Trust Principles., opens new tab

Huawei could seize China's AI chip crown in 2026 as Nvidia's H200 shipments stall in regulatory limbo -- Beijing pushes homegrown AI hardware dominance in a market projected to hit $67 billion by 2030

A Financial Times report has claimed that Shenzhen-based Huawei is on track to capture the largest share of China's AI chip market this year, following growing demand from Chinese firms seeking domestic alternatives to American chipmaker Nvidia. The report comes as Huawei reportedly expects AI chip revenue to surge to $12bn, up from $7.5bn in 2025. The company is basing its forecast of a 60% surge on existing orders for its 950PR chip, which entered mass production just last month. Huawei plans to launch an upgraded version of 950DT in the fourth quarter, as it continues to aggressively expand its chipmaking capabilities. The move comes as NVIDIA's China operations -- once accounting for up to 25% of its data center business revenue -- are being affected by export restrictions and regulatory barriers imposed by both the United States and China amid broader technological and trade tensions between the two countries. We reported earlier that Nvidia CEO Jensen Huang confirmed in March 2026 that the company had received U.S. licenses to sell H200 AI chips to China and was restarting production to meet demand. However, despite obtaining U.S. clearance and securing orders from Chinese customers, shipments have faced hurdles, with reports suggesting potential delays due to Chinese import regulations. The Financial Times report claims that Beijing has instructed Chinese tech companies to limit their use of Nvidia chips to their overseas operations, while supporting domestic manufacturing. On the other hand, US regulators require that all NNvidia chips ordered by Chinese clients only be used in China. These contradictory demands have led to a stalemate in customs clearance for H200 shipments to China. Most of Huawei's AI chips are manufactured by Semiconductor Manufacturing International Corporation (SMIC), China's leading fab. The company plans to add two additional dedicated fabrication plants this year. If it successfully ramps production, then its initial revenue forecast will likely see a boost. In all these, Nvidia's chips remain more advanced. However, Huawei is taking a different strategic approach to compete with Nvidia in AI hardware. The company has improved its chips enough to target a rapidly growing part of the AI market: inference -- the computation AI models use to generate answers and perform real-world tasks after training is complete. According to the Financial Times, the Chinese tech group has positioned its latest 950PR processors as the hardware of choice for domestic companies running inference. Huawei believes inference will become the biggest source of AI computing demand as AI assistants and autonomous agents become more common. Since inference workloads are generally less demanding than training massive AI models, Huawei can remain competitive despite weaker raw chip performance. To compensate further, Huawei is linking large numbers of its chips together using its networking technology to create powerful AI computing clusters, effectively boosting overall system performance even if each individual chip is weaker. The gamble appears to be working, starting with DeepSeek. While its latest v4 model was trained on Nvidia chips, DeepSeek used Huawei's 950PR for inference, the company confirmed last month. Nvidia apparently sees this as concerning. "The day that DeepSeek comes out on Huawei first, that is a horrible outcome for our nation - It could lead to a scenario where AI models around the world are developed and they run best on non-American hardware," said Nvidia's Huang in a recent interview with podcaster Dwarkesh Patel. For now, Nvidia still holds several major advantages. One of the biggest is its software ecosystem. While Huawei has been developing its Cann platform as a domestic alternative to Nvidia's widely used CUDA software, developers say it still lags significantly behind in usability and maturity. The difficulties involved in working with Cann can increase development complexity and operating costs for customers, helping NVIDIA maintain a strong grip on the global AI ecosystem. Regardless, Nvidia's grip on the Chinese AI chip market has weakened considerably. The sector was once dominated by the American chipmaker, largely through sales of its H20 processor -- a modified version designed specifically for the Chinese market to comply with U.S. export restrictions. However, after Washington moved to block sales of the H20 to China last year, Nvidia has struggled to find a replacement product that satisfies both American export rules and Chinese regulatory requirements. According to Morgan Stanley, China's AI chip market could grow to roughly $67 billion by 2030, with domestic companies expected to supply around 86% of that demand. The firm also estimates that Chinese suppliers alone could account for about $21 billion of the market this year, highlighting how rapidly the country's homegrown AI semiconductor industry is expanding. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Nvidia Reports Zero China AI GPU Share Amid Export Restrictions Shift

Nvidia CEO Jensen Huang has stated that the company now holds effectively no share in China's AI accelerator market, a significant reversal from its previously dominant position. The disclosure highlights the growing impact of U.S. export controls, which have restricted the company's ability to supply advanced AI hardware into one of the world's largest technology markets. According to Huang, Nvidia's decline in China is directly tied to these regulatory measures, which limit the export of high-performance GPUs used in AI workloads. While earlier projections suggested a steep drop in market share, the situation has progressed further, with Nvidia no longer participating in direct sales within the region. This has created a vacuum that domestic Chinese vendors are rapidly moving to fill. Companies in China are scaling up development of local alternatives, covering both hardware and software layers of the AI stack. With strong government backing, access to a large engineering workforce, and comparatively lower energy costs, China is positioned to expand its internal capabilities at pace. Analysts expect domestic suppliers to meet a substantial portion of demand, reducing reliance on foreign technology providers. Huang also raised concerns about the strategic implications of conceding a market of this size. He indicated that limiting U.S. companies' participation may accelerate China's transition toward self-sufficiency, potentially weakening the global reach of American-developed AI ecosystems. While restrictions may have been justified initially, he suggested that such policies need continuous reassessment in response to evolving market conditions. The Nvidia CEO further emphasized that China remains a competitive force in AI development, even without access to leading U.S. hardware. Its depth of technical talent and scale of research activity continue to support rapid progress in frontier models. In this context, the absence of U.S. firms in the region may influence long-term competitive dynamics rather than simply reshaping short-term market share. Looking ahead, the situation underscores a broader shift in the global AI industry, where regulatory policy, supply constraints, and regional innovation are increasingly interconnected. Nvidia's position in China serves as a case study of how geopolitical decisions can directly affect technology adoption, vendor presence, and ecosystem development at scale. Source: Special Competitive Studies Project via Tom's Hardware

NVIDIA CEO Jensen Huang says the company now has 0% market share in China

TL;DR: Due to strict US restrictions on AI-related hardware sales, NVIDIA's direct market share in China has dropped to zero, prompting Chinese technologies to fill the gap. CEO Jensen Huang criticizes this policy, emphasizing China's growing AI talent and self-sufficiency despite limited access to NVIDIA's GPUs and software. With strict restrictions on what US-based hardware can be sold in China for GPUs and other AI-related technologies, the direct result, according to NVIDIA CEO Jensen Huang, is that the company's market share in the region has dropped to zero. "In China, we have now dropped to zero," Jensen Huang said in a recent interview with the Special Competitive Studies Project. Naturally, with the boom in AI accelerator and data center equipment sales, a lot of money and market share are being left on the table. And yeah, Jensen Huang isn't happy with the situation, adding that it doesn't make "strategic sense" and that it has "already backfired" as Chinese-based technologies fill the void. "Conceding an entire market the size of China probably does not make a lot of strategic sense, so I think that has already largely backfired," Jensen Huang continues. "Maybe it made sense at the time, but I think the policy really needs to be dynamic and needs to stay with the times. I think it would be fairly safe to say that having American chip companies and other companies in China makes a lot of sense." Now, this isn't to say there aren't any NVIDIA GPUs in China being used for AI; the "zero" here refers to NVIDIA's direct sales to Chinese companies. Jensen Huang's argument that NVIDIA hardware should be available in China extends to software, CUDA, and other layers. And with these restrictions, China's AI endeavors are not only flourishing but are quickly becoming self-sufficient and competitive. "The argument there is that across the five-layer cake, there's one particular layer that is too important because in the others, China can get ahead," Jensen Huang continues. "They have cheaper energy. They have incredible talent. So, they have a number of science and math experts, and as a result of that, the number of AI researchers in China is quite extraordinary, it's one of their national treasures."

​Beyond China: Nvidia bets big on the global AI boom​ - ​Nvidia's China Shock and The Bigger AI Reality

Nvidia has witnessed a dramatic collapse in its China business, with CEO Jensen Huang confirming that the company now holds virtually zero market share in the country. This is a sharp reversal from its earlier dominance of nearly 95% in China's AI chip market, making it one of the most striking pullbacks by a global tech leader. The collapse has largely been driven by U.S. export restrictions that prevent Nvidia from selling its most advanced AI chips to Chinese customers. These regulatory constraints made its offerings less competitive, while Chinese companies quickly shifted to domestic alternatives. As a result, Nvidia has effectively lost access to one of the world's largest AI markets.

NVIDIA's CEO Jensen Huang Draws a Hard Line on China: No Blackwell, No Rubin, But US Firms Must Still Fight In Global Markets

Blackwell or Rubin AI Chips are a no-go for China, but NVIDIA's CEO Wants US Firms To Continue To Compete In Global Markets The US AI policy shift towards China has been ongoing since NVIDIA's Hopper generation of chips. Export controls restricted NVIDIA from selling its bleeding-edge chips to Chinese firms & this ban has extended into the Blackwell generation. Reaffirming the US-first policy, NVIDIA CEO Jensen Huang has stated that the company's most advanced AI chips, Blackwell and Rubin, won't be given to China. Jensen says that America should always have a head start in the AI chips industry because NVIDIA is an American company & its latest AI technologies, whether hardware or software, are made for the US first. Becky Quick (Anchor "Squawk Box"): I understand your point that we need to be engaging with other nations. And I believe you're probably referring to China, specifically, that we need to be able to engage with them. And I know your position has been that we should be giving them the H200 chips. Not necessarily the latest and greatest, but we should be giving them chips so that they are reliant on US companies for some of these things. That makes sense. Jensen Huang (NVIDIA CEO): We should compete globally. America should always have Head Start. Becky Quick (Anchor "Squawk Box"): Should they have the latest and greatest chips? Jensen Huang (NVIDIA CEO): No. We have the right, the United States, we're an American company, the United States has the right to make sure that, and we're delighted by that, and we're huge supporters of it, that the United States has the first, the most, and the best. But simultaneously, all American companies should compete globally. Because remember, in the final analysis, we're trying to maximize exports. We're trying to maximize American exports. We're trying to increase our revenues. And by increasing our revenues, tax revenues, we improve our national security. And economic security contributes to national security. Tax dollars helps us with defense. All of that increases national security. American technology has to win across the world at every single layer. During the "Leading in the Age of AI" talks at the Miliken Institute's Global conference, Jesnen also pointed out that while the latest chips are built for the US soil first, US firms should compete globally. While the United States gets "the first, the most, and the best" of advancements from American firms, the companies should also be open to competing globally, as that helps increase revenues, which in turn helps improve the United States' National Security. This statement comes just a few weeks after NVIDIA's CEO announced that they have zero share in China. We recently reported how NVIDIA's "official" share in China has dropped to nothing as China itself is focusing on in-house solutions, and companies such as Huawei are now gaining big time for their growing AI demand. Jensen also confirmed that they have not shipped a single unit of their Hopper H200 GPUs to China. That is despite NVIDIA being given the go-ahead to sell H200 chips to China after paying a 25% of the sales to the US Government. The President of the US, Donald J. Trump, has already stated that NVIDIA's "SUPER DUPER" Blackwell GPUs are the most advanced AI chips designed by NVIDIA, and there might be a chance that a cut-down variant of the chip, with severely reduced capabilities, can be offered to China, though those plans haven't seen any progress.

Huawei Is The Biggest Winner In China's AI Market After NVIDIA Pullout, AI Share To Reach 60% This Year

NVIDIA pulling out from China's AI market has boosted the share of domestic firms, with Huawei winning the biggest chunk. The US Government has moved to ban all leading-edge AI chip sales in China. NVIDIA, being the biggest name in the AI industry, has seen its share drop to zero after the policy shift, prompting an increased reliance on domestically produced chips in China. Currently, the situation has prompted China's AI chipmakers to double down on production and fill in the gaps created by NVIDIA's departure in a significant way. While NVIDIA still has older chips for China, these aren't viable solutions anymore for emerging AI workflows such as Agentic AI. The entire AI economy now shifts to in-house solutions. Huawei is the leading AI and datacenter chipmaker in China, and given these restrictions on NVIDIA GPUs, the company is all set to grab a 60% share in Chinese AI markets by the end of 2026. Previously, Huawei was the only brand that could've tackled NVIDIA in the cost/$ segment, and their roadmap, being robust as ever, makes them a formidable player, which, without any imminent competition, can become the single and most powerful AI entity in the country. Huawei is set to capture the largest share of China's AI chip market this year, with sales jumping by at least 60 per cent amid strong demand from Chinese companies seeking domestic alternatives to Nvidia. via Financial Times The current figures estimate that Huawei's AI chip revenue could hit $12 billion in 2026, based on current orders. The bulk of this revenue is made from orders of the Ascend 950PR chip, which entered mass production in March. The Ascend 950PR features support for MXFP4, a key instruction set that drives AI models such as DeepSeek V4, and features up to 2 PFLOPs of FP4 performance, along with 128 GB of locally produced HBM memory. An updated version of the Ascend 950 lineup, called 950DT, is also expected to enter mass production in Q4 2026, featuring 144 GB of HBM memory. The company has the Ascend 960 planned for Q4 2027 and the Ascend 970 planned for Q4 2028. Each chip offers substantial updates in performance, new features & increased HBM memory. Morgan Stanley forecasts China's AI chip market will reach $67bn in 2030, with 86 per cent expected to be supplied by Chinese players. It estimates the market to be about $21bn this year from domestic suppliers. via Financial Times DeepSeek's recent V4 announcement saw both NVIDIA and Huawei offering Day-0 support, marking a huge win for the Chinese AI semiconductor markets. Huawei relies mostly on SIMC for its chip production, and the US is cracking down on advanced lithography machinery sales from ASML to China. Despite these blocks, China continues its AI infrastructure and software development at a steady pace.

Big Chinese tech firms scramble to secure Huawei AI chips after DeepSeek V4 launch, sources say

BEIJING, April 29 (Reuters) - Demand for Huawei's Ascend 950 AI chips has surged following the release of DeepSeek's V4 artificial intelligence model that runs on the Shenzhen-based tech firm's chips, with major Chinese internet firms rushing to secure orders, three people familiar with the matter said. China's biggest internet firms including ByteDance, Tencent and Alibaba are reaching out to Huawei about new chip orders, said the sources, who are familiar with the procurement discussions. Companies specializing in cloud computing and graphics processing unit (GPU) rental services are also scrambling to place orders, two of the sources added, without providing the names of the firms. While the 950PR significantly outperforms Nvidia's H20 chip - the most powerful chip Nvidia was permitted to sell in China until Beijing blocked its import last year - it still trails the American firm's H200, a more advanced processor that has been caught up in regulatory limbo. Despite U.S. and Chinese approvals for exports, the H200 has yet to be shipped to China as Beijing and Washington remain at odds over the conditions governing its sale, providing an opportunity for Huawei to sell its semiconductors. The 950PR represents a breakthrough for Huawei after years of struggling to win large orders from China's tech sector. Customer testing of the chip went well earlier this year, with firms including ByteDance and Alibaba planning orders after samples were distributed in January, Reuters reported in March. Huawei, ByteDance, Alibaba and Tencent did not respond to Reuters requests for comments. DEEPSEEK FRENZY The scramble for Huawei's chips underscores how DeepSeek's V4 release last week has turbocharged demand for domestic Chinese AI hardware as U.S. export controls continue to restrict access to Nvidia's most advanced processors. It is also an endorsement of the performance of Huawei's chips so far. DeepSeek's decision to optimize its V4 specifically for Huawei's chips marks a strategic shift away from American semiconductor dependence and more towards China's homegrown AI gear, which is a priority for Beijing as it seeks tech supremacy. Last week, Huawei said its Ascend supernode infrastructure, built on the Ascend 950 series chips, would fully support DeepSeek V4 models and that the entire Ascend SuperNode product line had been adapted for V4 inference, referring to the process of using a trained AI model to answer queries and execute tasks. Among Chinese chipmakers, Huawei's Ascend 950 series - specifically the 950PR variant - is the only domestic chip to support a technique that processes AI calculations in a more compressed numerical format, allowing it to handle more computations per second at a lower cost. Highlighting the rush for access, Alibaba Cloud's Bailian platform made DeepSeek V4 available on the day that it was released, offering both the V4-Pro and V4-Flash variants with pricing matching DeepSeek's official rates. Tencent Cloud launched V4 preview services on its TokenHub platform on the same day, deploying the model on both domestic nodes and its Singapore international gateway to serve global users. The rapid deployment by major cloud platforms means millions of users and developers can now access V4, sharply increasing the volume of AI queries that need to be processed - and with it, demand for the underlying chips. SUPPLY CONSTRAINTS PERSIST DeepSeek, which is offering developers a 75% discount on its new model until May 5, said V4-Pro pricing could decline significantly in the second half of 2026 once Huawei's Ascend 950 supernodes "ship at scale." However, the company acknowledged that constraints would persist until production ramps up, reflecting the tight supply of high-end homemade AI chips. DeepSeek's V4 includes two versions: V4-Pro with 1.6 trillion total parameters and V4-Flash with 284 billion parameters, both supporting a one-million-token context window. The models are available as open-source releases under the permissive MIT open-source licence, which allows companies to freely use, modify and commercialise the models. However, output of the 950 is expected to fall short of demand due to U.S. export restrictions on advanced chipmaking tools that prevent China from acquiring cutting-edge manufacturing equipment. Huawei planned to ship around 750,000 units of the 950PR this year, with mass production beginning in April and full-scale shipments starting in the second half of 2026, according to people familiar with the plans. (Reporting by Beijing and Hong Kong newsrooms; Editing by Anne Marie Roantree and Thomas Derpinghaus)