For years, the U.S.βChina chip conflict was described as a restriction: a set of rules limiting what could be sold. By October 2026, it is better described as a market structure. Export controls did not stop Chinese AI development. They reorganised it β into a parallel market with its own suppliers, its own product tiers, and its own scarcity economics. Two October reports, both drawing on Reuters reporting and IDC data, show how far that reorganisation has gone.
First, Nvidia's share of China's market for data-centre AI chips fell to 34% in the first half of 2026, according to an IDC report seen by Reuters and published October 13, 2026. Nvidia shipped 678,000 AI accelerators in China from January through June, down 44% from a year earlier. Huawei shipped 572,000 accelerators, up from more than 300,000 a year earlier, for a 28% share. Alibaba's chip-design unit T-Head held 7%, followed by Cambricon with 5%, Baidu's Kunlunxin with 4% and Hygon with 3%. Nine of the 11 Chinese chipmakers tracked by IDC increased shipments year on year. Total AI-accelerator shipments reached 2.008 million units in the half, even as GPU shipments fell 26% to 991,000 β a sign that domestic supply, not demand, is the binding constraint.
The $35 Billion Order That Reveals the Scarcity
Second, Reuters reported on October 20, 2026 that ByteDance, Alibaba and Tencent have ordered more than 3 million of Nvidia's new RTX PRO 5500 workstation graphics cards, a batch with a list price above $35 billion, citing two people familiar with the matter. At roughly $11,400 apiece, the cards are a less powerful but export-compliant option for Chinese companies shut out of Nvidia's best hardware. The Blackwell-based B300 data-centre chip is barred from sale to Chinese buyers, and the Hopper-based H200 can be exported only after case-by-case review under licensing conditions. The RTX PRO 5500 is a workstation card, not a data-centre chip β yet it ranks among the strongest Nvidia products Chinese customers can still buy, and the companies want it mainly for AI inference, running models already trained.
The scarcity is stark. Total orders from customers in China have exceeded 6 million cards, according to Reuters' sources, while Nvidia told partners they might receive only 5% to 10% of the quantities requested, and indicated it could produce no more than 3 million RTX PRO 5500 cards globally in 2027. Nvidia has told customers to expect early samples in December 2026, with commercial deliveries beginning in the first quarter of 2027. The predecessor card, the RTX 6000D, has been discontinued, and one source said its market price in China has doubled. A Shenzhen-listed firm, Tansun Technology, disclosed in an October 13 filing that it had agreed to buy 20,000 RTX PRO 5500 cards from an unnamed seller. This is a shortage market inside a restriction regime: even compliant products cannot be had at the volumes buyers want.
What Decoupling Actually Means for Builders
Nvidia CEO Jensen Huang has repeatedly criticised the export curbs, saying the policy forced a U.S. company to cede China's $50 billion AI chip market to Huawei, according to the Reuters report. Whether or not one agrees with that framing, the IDC data supports its direction: the share loss is real, the domestic gain is broad-based across many suppliers rather than one champion, and the demand for inference capacity β the cheaper, higher-volume end of AI compute β is pulling whatever compliant supply exists into China at extraordinary prices.
For founders, three implications follow. First, any hardware or AI-infrastructure company with a China-adjacent supply chain must plan for two product tiers: a frontier tier that cannot legally ship to China, and a compliant tier whose allocation is uncertain. Product roadmaps that assume one global SKU are obsolete. Second, the rise of domestic Chinese accelerators creates a second software ecosystem: models, frameworks, and optimisation tools tuned for Huawei, Cambricon, and T-Head hardware will improve with volume, and companies selling into Asian markets will increasingly need to support them. Third, scarcity itself is a business model: allocation, brokerage, and secondary-market pricing for compliant chips are already visible in the RTX 6000D's doubled price.
What It Means for Founders and Investors
For investors, the China chip market is no longer a question of "when do restrictions lift?" but "how do you underwrite a permanently bifurcated market?" Companies whose value depends on unfettered access to the best American hardware in China carry a policy risk that no technical moat can hedge. Companies that build for the compliant tier β inference-optimised software, domestic-accelerator tooling, or allocation-efficient infrastructure β are building inside the market as it actually exists.
The deeper lesson is about industrial policy as market design. Export controls were intended to slow a rival. They also created, inadvertently, the conditions for that rival's domestic industry to scale: guaranteed demand, protected pricing, and a clear performance target. Founders and investors operating anywhere near this fault line should treat geopolitics not as background noise but as a first-order variable in product, pricing, and market-entry decisions.
Angeled.com connects founders with investors.
Reuters (via WNCY), "Nvidia's China AI-chip share falls as Huawei closes gap, report finds" (October 12, 2026)
Source: fourweekmba.com
FourWeekMBA, "ByteDance, Alibaba, Tencent Order $35B of Nvidia Chips" (October 2026)
Source: whatledto.com
WhatLedTo, "US Export Controls on Nvidia AI Chips to China β Timeline" (updated October 25, 2026)
Source: medium.com
Medium, "Chinese Tech Giants Line Up for $35 Billion of Nvidia's Export-Compliant AI Chips" (October 2026)
Read Next
Open Weights Are Not Open Business Models: Mistral Large 4 and the Licensing Trap
Courts Are Writing AI's Rules Now: Three Rulings Founders Should Read
Delhi High Court Orders 24-Hour Takedown of AI Deepfake Videos Targeting Chief Minister
EU Court Adviser Says AI Training Is Legal β but Gemini's Answers May Still Infringe Copyright
AI Regulation After the Midterms: What Founders Should Do Before the January Bills Arrive
DOJ Launches Super Intelligence Task Force to Police AI Misuse



