Hua Hong Semiconductor and SMIC just showed you where China's chip fight has moved: not to a clean win over Nvidia, but to the crowded mature-node fabs feeding the AI buildout around it.
Hua Hong Semiconductor's second quarter was not a small rebound. It was a hard turn. The company reported net profit attributable to owners of $38.6 million for the quarter ended June 30, 2026, up 385.9% from a year earlier and 84.6% from the first quarter. Revenue hit a record $717.5 million, up 26.8% year over year. Gross margin rose too: to 16.5%, from 10.9% a year earlier.
That beat Hua Hong's own May guidance, which had pointed to second-quarter revenue of $690 million to $700 million and gross margin of 14% to 16%. The year-earlier comparison is important, because Hua Hong didn't need a story about distant breakthroughs to explain the jump. It shipped more wafers, charged better prices, and kept its factories busy. Wafer shipments rose 17.9% year over year to 1.538 million 8-inch equivalent wafers, according to the company's results.
SMIC, China's largest foundry, went further. The Wall Street Journal reported that SMIC's second-quarter net profit more than tripled to $479.2 million, while revenue rose 36% to $3.01 billion and gross margin climbed to 25.3%. Its utilization rate reached 93.7%. For a company still boxed in by US controls on leading-edge tools and chips, that is a blunt result.
The demand Washington didn't block #
The important fact is not that either company has caught TSMC at the cutting edge. It hasn't. Hua Hong's strength is in specialty and mature processes, and SMIC still sits well behind the most advanced Taiwanese production. But AI systems don't run on accelerators alone. They need power-management chips, image sensors, radio-frequency components, controllers, networking parts, you name it. That layer is less glamorous than an H200 or an H100. It also ships in volume.
Europe bets €10 billion on AI gigafactories but Nvidia still holds the keys The European Commission launched tenders today for seven AI gigafactories backed by €10 billion in public funding, part of a €30 billion push to give European AI startups sovereign compute access. But with Nvidia supplying the chips and US hyperscalers holding 70 percent of the EU cloud market, the sovereignty question is more complicated than the... - European AI chip manufacturing strategy - Nvidia dependency Europe computing sovereignty
That's where the money is showing up. SMIC has pointed to higher pricing on short-supply products and rising demand for power-management and peripheral chips tied to the AI boom. Hua Hong has seen the same broad pull through its specialty platforms. If you're only watching whether China can replace Nvidia's best accelerator, you're missing the more immediate shift: Chinese foundries are becoming the place where the surrounding AI hardware stack gets built.
Reuters reported in April, citing IDC data, that Chinese GPU and AI chip makers captured nearly 41% of China's AI accelerator server market in 2025, while Nvidia still led with about 55%. That was already a sharp drop from Nvidia's roughly 95% share before sanctions. South China Morning Post later cited TrendForce forecasts showing domestic suppliers led by Huawei and Cambricon set to take a much larger share of China's AI server market in 2026. The direction is clear. Uncertainty pushed customers to local options, and once supply chains move, they don't politely move back on Washington's schedule.
Hua Hong is buying capacity while it can #
Hua Hong is not just enjoying the cycle. It is using it. In July, China's securities regulator approved Hua Hong Grace's plan to buy 97.4988% of Shanghai Huali Microelectronics through a share issue, a transaction priced at 8.268 billion yuan, or roughly $1.15 billion. Huali brings 12-inch foundry lines focused on 65/55-nanometer and 40-nanometer processes, exactly the kind of mature and specialty capacity now benefiting from AI-related demand.
The timing matters. The US Bureau of Industry and Security eased part of the AI chip regime in January 2026 by moving Nvidia's H200, AMD's MI325X and similar chips to case-by-case review for China and Macau, subject to strict conditions. A separate presidential proclamation put a 25% tariff structure around covered advanced computing chips, and policy analysts at CNAS noted that shipments to China are capped against US domestic sales. On paper, that gives Nvidia and AMD a controlled path back into China.
In practice, it doesn't unwind three years of planning around risk. Chinese cloud firms and chip designers have had every reason to avoid building roadmaps around a license that can change with the next rule. Huawei's Ascend line, Cambricon, SMIC, Hua Hong and other domestic suppliers don't need to win every benchmark to benefit from that fear. They only need to be available when the purchase order is due.
None of this makes China's chip sector self-sufficient. It still faces serious limits in lithography, high-bandwidth memory, advanced packaging and top-end GPU software. Frankly, pretending otherwise is just nationalist marketing in another form. But the second-quarter numbers from Hua Hong and SMIC show something harder for Washington to dismiss: export controls can slow one layer of the stack while pushing demand into another.
Hua Hong guided for third-quarter revenue of $770 million to $780 million, with gross margin between 16% and 18%. SMIC forecast a further 2% to 4% revenue increase and gross margin of 26% to 28%. Those are not victory laps. They are factory schedules. For now, the mature-node business around AI is doing exactly what policy pressure encouraged it to do: stay in China and scale.
Nvidia is now bankrolling the company that buys its chips and the numbers are getting hard to ignore
Nvidia is in talks to guarantee up to $250 billion for OpenAI's data center buildout in Ohio, with separate discussions covering $350 billion more in chip financing. The deal reignites circular-financing concerns as Nvidia's total AI deal exposure crossed $750 billion last week. - nvidia financing openai infrastructure deals - gpu supplier funding data center buyers
Also read: Dodgers Owner Mark Walter's Insurer Will Shed $6.5 Billion in Self-Dealing • Kalshi Must Geofence Washington Bettors Today or Face Steep Daily Fines • Accounting startup Rillet hits $1 billion valuation as AI eats CFO busywork