Cambricon Technologies reported Q1 2026 revenue of 2.88 billion yuan ($421 million), up 160% year-on-year, with net profit of 1.01 billion yuan – up 185% from the same period last year. The company posted its first-ever full-year profit in 2025, with annual revenue of 6.5 billion yuan compared to 1.2 billion yuan in 2024, and net income of 2.1 billion yuan against a 452 million yuan loss. Shares surged 14% in Shanghai on the Q1 filing day. Cambricon is targeting delivery of 500,000 AI accelerators in 2026, up from an estimated 116,000 in 2025, with the advanced Siyuan 590 and 690 chips at the centre of that production ramp. YourNewsClub views the Q1 revenue figure – more than twice the entire 2024 annual total in a single quarter – as the most operationally concrete measure of how rapidly Chinese AI infrastructure investment is translating into demand for domestic chips.
The driver is policy, not purely technology. Beijing has actively encouraged China’s leading AI developers – including ByteDance, Alibaba, and Tencent – to prioritise domestically produced AI accelerators given ongoing restrictions on Nvidia’s most advanced chips. Cambricon is one of only two companies on China’s government-approved AI hardware procurement list; the other is Huawei. ByteDance reportedly pre-ordered approximately 200,000 Siyuan 590 chips in 2025, making it the company’s dominant buyer at roughly 80% of revenue in the first half of that year. Whether that concentration eased in subsequent quarters is unclear from public disclosures. Nvidia’s share of the Chinese AI chip market has declined from a pre-sanctions peak of approximately 95% to below 60% by early 2026, even as the US government has approved limited H200 exports. That gap – from 95% to under 60% – is the market that Cambricon, Huawei, MetaX, Moore Threads, and Biren Technology are competing to fill.
The supply-side constraint is SMIC. Cambricon relies on SMIC’s N+2 seven-nanometer-class process for its Siyuan chips. TrendForce has estimated SMIC’s yields at that node at around 20% for the most advanced configurations, and SMIC’s capacity expansion is constrained by restrictions on importing critical semiconductor equipment. That yield ceiling and capacity bottleneck are the primary limits on how fast Cambricon can actually deliver the 500,000 units it is targeting. The Siyuan 690 next-generation chip is still in testing, with large-scale production potentially slipping to the second half of 2026. Morgan Stanley estimates the Chinese market for AI chips could reach $67 billion by 2030. At current growth rates, Cambricon’s share of a market that large would imply a business significantly bigger than today’s already rapid expansion. YourNewsClub rates the SMIC yield and capacity constraint as the most commercially binding variable in whether Cambricon’s 500,000-unit target becomes a delivery number or an aspiration.
Maya Renn, whose work focuses on the ethics of computation and access to power through technology, frames what concentrated procurement means for the market structure: “When 80% of a company’s revenue comes from one customer and that customer is also a state-policy-aligned buyer, the commercial growth story and the geopolitical story are inseparable. Cambricon’s Q1 results are an accurate measure of Chinese AI infrastructure investment – they are also a measure of how effectively Beijing has redirected procurement away from American technology.” Freddy Camacho, who studies the political economy of computation and capital as dominance assets, places the capital cycle: “Chinese AI chipmakers are experiencing the same dynamic as their American counterparts in 2020: government policy, concentrated infrastructure investment, and a captive procurement environment producing revenue growth that looks like commercial success but is primarily state-demand-driven. The long-term question is whether that revenue base survives when the policy subsidy normalises.”
Your News Club tracks the H2 2026 Siyuan 690 production ramp as the most commercially revealing near-term data point: if Cambricon ships at scale on its advanced node in the second half of 2026, it establishes a technology position that competitors including Huawei and MetaX will need to match.
YourNewsClub counts the SMIC yield rate at the N+2 node – estimated at approximately 20% for the most advanced configurations – as the single most constraining production variable in whether the 500,000-unit target becomes a delivery figure or a plan.