Huawei and Cambricon Tighten Grip on China’s AI Chip Market

Chinese domestic chip suppliers, led by Huawei Technologies and Cambricon Technologies, are on track to capture nearly 80% of China’s AI server chip market in 2026, according to Taipei-based research firm TrendForce. The dramatic shift marks a structural transformation in one of the world’s most strategically important technology markets, as US export controls and Beijing’s push for semiconductor self-sufficiency accelerate Nvidia’s retreat.

Foreign suppliers, including Nvidia and AMD, are projected to see their combined share of China’s AI server chip market fall from 34% in 2025 to just 21% in 2026, according to data presented by TrendForce research manager Frank Kung at a conference in Shenzhen. Domestic chip suppliers led by Huawei and Cambricon are expected to increase their share from 46% to 56%, while Chinese internet companies’ custom application-specific integrated circuits (ASICs) will grow from 20% to 23%.

“This segment will continue to scale up,” Kung said, as geopolitical uncertainty and Beijing’s push for technological self-reliance accelerate adoption, as reported by the South China Morning Post.

Huawei’s Ascend Dominance

Huawei has emerged as the clear leader in China’s domestic AI chip market. According to Bernstein Research, Huawei’s AI chip market share in China reached 40% in 2025, tied with Nvidia. By 2026, Huawei is projected to reach 50% while Nvidia’s share shrinks to just 8%, as reported by Sina Finance.

In terms of actual shipments, China’s total AI accelerator card deliveries reached 4 million units in 2025. Domestic chip deliveries hit 1.65 million units, capturing 41% market share. Huawei’s Ascend series alone shipped 812,000 units — half of the domestic total. Alibaba’s T-Head shipped 265,000 units, while Baidu’s Kunlun Core and Cambricon each shipped 116,000 units, according to AI Chip Front.

Cambricon’s Financial Turnaround

Cambricon Technologies achieved 2025 revenue of 6.497 billion RMB (approximately $900 million USD), a year-over-year increase of over 450%, marking its first profitable year since listing five years ago. The milestone underscores that domestic AI chip companies have begun to establish viable commercial pathways from research and development to scaled profitability.

DeepSeek Migration Signals Ecosystem Shift

In a landmark move for China’s AI ecosystem, the Chinese AI model leader DeepSeek announced that its next-generation flagship model would be built on Huawei’s Ascend 950PR computing base, abandoning Nvidia’s CUDA framework for Huawei’s proprietary CANN platform. The decision represents a significant validation of Huawei’s software ecosystem and signals a broader migration away from Nvidia’s industry-standard platform.

Policy Drivers Accelerating the Shift

Government mandates are playing a critical role in accelerating domestic chip adoption. Chinese cities including Beijing and Shanghai have mandated that 50-100% of new smart computing center chips must be domestically produced. “Computing vouchers” — government subsidies designed to reduce trial costs — are being deployed to encourage companies to adopt domestic alternatives.

On the demand side, China’s daily token calls surpassed 140 trillion in Q1 2026, driving demand for cost-effective inference chips over expensive training chips. This shift in demand patterns favors domestic suppliers who can offer competitive pricing and adequate performance for inference workloads.

Global Context: Nvidia Still Dominates Worldwide

Despite losing ground in China, Nvidia is projected to maintain approximately 64% of the global AI chip market in 2026, with AMD at 8.6%, according to TrendForce. The bifurcation of the global AI chip market into two distinct ecosystems — one centered on Nvidia’s CUDA in Western markets and one centered on Huawei’s CANN and other domestic platforms in China — represents one of the most significant technological consequences of the US-China technology conflict.

Implications and Outlook

The structural reversal of China’s AI chip market — from Nvidia-dominated to domestic-dominated in approximately three to four years — has far-reaching implications. For Nvidia, the loss of the Chinese market represents a substantial revenue hit, even as global demand remains strong. For Chinese tech giants like Alibaba and Baidu, the line between chip designer and chip consumer is increasingly blurring as they develop custom silicon.

Looking ahead, key questions remain: Can Huawei and Cambricon maintain their performance trajectory and close the gap with Nvidia’s latest architectures? Will Chinese companies face new bottlenecks in advanced manufacturing, particularly regarding EUV lithography access? And how will US export control policy evolve in response to China’s domestic chip success?

What is clear is that China’s AI chip market has undergone a fundamental transformation. The era of Nvidia dominance in China is over, and a new chapter of domestic-led innovation — and global technological fragmentation — has begun.