China’s most consequential semiconductor project is scaling rapidly. Huawei Technologies, the telecommunications and technology giant operating under a comprehensive set of U.S. export restrictions, is targeting the production of approximately 600,000 units of its marquee Ascend 910C artificial intelligence accelerator in 2026.
The production target, originally reported by Bloomberg, has come into sharp focus this week as China’s domestic AI chip sector enters a landmark earnings season, with multiple companies posting triple-digit revenue growth driven by a state-mandated campaign to replace foreign hardware.
The 910C is Huawei’s most advanced AI chip, designed to serve as a domestic alternative to Nvidia’s restricted H100 and H800 accelerators. While the Ascend chip trails Nvidia’s best products in raw single-unit performance, a gap attributable in large part to China’s inability to access extreme ultraviolet lithography equipment from Dutch manufacturer ASML, Huawei has compensated through system-level engineering.
At China’s World AI Conference in Shanghai in July, the company showcased massive server racks capable of integrating tens of thousands of Ascend chips, demonstrating that scale and interconnect architecture can partially offset the per-chip performance deficit.
The Import Substitution Mandate
The push to produce 600,000 Ascend 910C units is not merely a commercial decision, it is a response to a direct government mandate. Beijing has ordered local technology firms to find alternatives to Nvidia’s cutting-edge products, accelerating a procurement shift that was already underway due to U.S. export controls. According to a June 2026 survey of Chinese technology executives by Bloomberg Intelligence, companies now expect to spend 46 percent of their AI accelerator budgets on locally made chips over the next 12 months, up sharply from the current 30 percent allocation.
The U.S. government did recently clear sales of Nvidia’s H200 chips, the most powerful AI processors ever permitted for export to China — to a limited number of domestic firms. However, American officials acknowledged last month that the actual number of Nvidia processors shipped to China since that authorization has been “trivial.”
The practical reality is that Nvidia and Advanced Micro Devices are virtually locked out of the Chinese market, as Bloomberg noted in its analysis of the sector’s earnings season, leaving Huawei and a handful of domestic designers to fill the enormous demand gap.
(Related: Huawei’s Top Chip Scientist Says Western Firms Are Approaching a Dead End)
Deployment Economics Over Peak Performance
The competitive dynamics of the Chinese AI chip market are shifting in ways that favor domestic suppliers. As Chinese enterprises move from the training phase, which demands the absolute highest computational throughput, to the inference and deployment phase, the economic calculus changes.
“Purchasing decisions are increasingly driven by deployment economics rather than absolute peak silicon performance,” Morgan Stanley analyst Charlie Chan wrote in a recent research note. In this environment, Huawei’s ability to offer reliable domestic supply, competitive token-generation pricing, and deep integration with Chinese software ecosystems becomes a significant advantage.
Morningstar analyst Phelix Lee offered a candid assessment of the market dynamics: “Most people would privately agree that having access to Nvidia would always be ideal, but in the absence of that, the trend I’m seeing is more Chinese chips being used.” This pragmatic shift in procurement behavior is reflected in the broader sector’s financial performance. Alongside Cambricon’s 108 percent revenue surge reported this week, Moore Threads projected first-half revenue growth of up to 149 percent, while Iluvatar CoreX is expected to see sales triple over the same period.
(Related: Samsung and SK Hynix Are Quietly Testing AMEC Chip Tools at Their China Fabs)
The Long-Term Self-Sufficiency Roadmap
The Ascend 910C production ramp is a critical milestone in China’s decade-long campaign to build a fully domestic AI hardware stack. Beijing’s ambitious data center buildout, which aims to spend 2 trillion yuan ($295 billion) over the next five years, is designed to rely on local suppliers for at least 80 percent of its core technology. Morgan Stanley now estimates that this structural shift will push China’s self-sufficiency level in AI chips to 70 percent by the end of the decade, a dramatic increase from 42 percent in 2025.
For Huawei, the 910C production target represents both a commercial opportunity and a national mission. The company’s ability to deliver hundreds of thousands of AI accelerators annually, despite operating under the most severe technology restrictions ever imposed on a major corporation, demonstrates the resilience of China’s semiconductor ecosystem.
Whether the 910C can ultimately close the performance gap with Nvidia’s next-generation Blackwell architecture remains an open question, but the sheer volume of chips being produced ensures that Chinese AI developers will have the raw compute necessary to continue training competitive frontier models.