A-share Hard Tech Q3 Reports: Computing Power Chain Delivers High Growth, Robotics Track Builds Momentum

Main body of the article

As A-share listed companies wrapped up disclosure of their 2026 third-quarter reports, the "hard tech" sector delivered a standout "report card": driven by both the global expansion of intelligent computing infrastructure and the localization of domestic computing power, the AI computing power industry chain's revenue and profit have maintained high growth, with industry prosperity remaining elevated, making it the main line with the fullest earnings realization this round; although the consumer electronics sector is under overall pressure from rising memory chip prices and weakening end demand, a group of companies is actively transforming and opening up a "second growth curve"; among frontier tracks, listed companies in the embodied intelligent robotics field have increased R&D investment and orders have surged, with the industry moving from proof of concept to the starting line of implementation.

Computing power sector delivers strong earnings

In reviewing this round of third-quarter reports, the reporter found that the computing power industry chain is the main line with the strongest earnings elasticity in the hard tech sector. The entire chain—from domestic AI chips to complete server systems and high-speed optical modules—has benefited from the global wave of AI infrastructure construction, and many companies' results significantly exceeded expectations. Xie Cun, director of the Information and Communications Administration of the Ministry of Industry and Information Technology, recently disclosed at a press conference of the State Council Information Office that as of the end of September 2026, China's intelligent computing power scale reached 2592 EFLOPS (100 quintillion floating-point operations per second), up 146% year on year, strongly supporting the rapid development of China's artificial intelligence industry.

Moore Threads, a leading domestic AI chip company, posted standout third-quarter results. Its financial report shows that the company's operating revenue in the third quarter of 2026 was 1.55 billion yuan, up 1773.9% year on year.

Cambricon, another domestic AI chip player, achieved operating revenue of 9.526 billion yuan in the first three quarters, up 106.76% year on year, and net profit attributable to shareholders of 3.715 billion yuan, up 131.49% year on year. According to Cambricon's restricted stock incentive plan, the 2026 revenue assessment target is 13.5 billion yuan, meaning about 3.974 billion yuan in revenue needs to be completed in the fourth quarter. The company's management said at an earnings briefing that demand in the AI computing power market remains robust and can support steady business growth.

On the complete-system side, Foxconn Industrial Internet's third-quarter report continued its high-growth trend, with single-quarter revenue reaching 383.554 billion yuan in the third quarter of 2026, up 57.73% year on year, and net profit attributable to shareholders of 16.634 billion yuan, up 60.35% year on year, marking five consecutive quarters of single-quarter net profit attributable to shareholders exceeding 10 billion yuan. The company said the core driver of its results was the continued explosion in AI computing power demand, with dual-line volume growth in cloud computing and networking businesses becoming the core force driving the company's continued strong growth.

Optical communications, a core link in computing power interconnection, shows notable volume and price increases across the industry. Zhongji Innolight's third-quarter revenue and net profit attributable to shareholders both hit record single-quarter highs, driven mainly by continued volume growth of 800G and 1.6T high-end optical modules; Eoptolink also said in its financial report that it has sufficient orders on hand, and order delivery and capacity expansion are proceeding according to business demand. The company is also making forward-looking arrangements in next-generation technologies such as NPO, CPO, LPO and silicon photonics, with NPO products expected to achieve mass shipments in 2027. Upstream optical component maker TFC Optical Communication and optical fiber maker Hengtong Optic-Electric also benefited from the upswing; Hengtong Optic-Electric's net profit attributable to shareholders in the first three quarters was 6.163 billion yuan, up 159.42% year on year, as the optical fiber market, driven by AI computing power, saw tight supply and demand, rising prices and higher gross margins.

Consumer electronics seeks new opportunities amid pressure

In contrast to the broad prosperity of the computing power industry chain, the consumer electronics sector's third-quarter reports showed clear divergence, with the core contradiction coming from the surge in memory chip prices suppressing end demand for smartphones and other terminals. An IDC (International Data Corporation) report shows that global smartphone shipments in the third quarter of 2026 were only 255.3 million units, down sharply 22% year on year, mainly because high component costs pushed up end prices and consumers delayed replacing their phones. IDC also said the industry has not yet bottomed out and forecasts that market recovery will not come until 2028 at the earliest.

Traditional consumer electronics leader Lens Technology came under earnings pressure, with revenue of 47.902 billion yuan in the first three quarters of 2026, down 10.74% year on year, and net profit attributable to shareholders of 1.42 billion yuan, down 50.06% year on year; third-quarter net profit attributable to shareholders fell 50.42% year on year. The company explained that revenue was affected by multiple factors including weak end demand, memory price increases suppressing smartphone and computer demand, renminbi appreciation, and the delayed pace of new products from major customers.

LCD panel leader BOE A also faced cyclical pressure on its traditional display main business. Its third-quarter report shows that BOE A achieved revenue of 48.677 billion yuan in the third quarter, down 8.62% year on year, and a net loss attributable to shareholders of 996 million yuan, turning from profit to loss year on year; in the first three quarters it achieved revenue of 151.809 billion yuan, down 1.77% year on year, and net profit attributable to shareholders of 4.251 billion yuan, down 7.61% year on year. However, leveraging its accumulated glass substrate processing capabilities, the company is extending into the AI advanced packaging track, and its glass-based packaging substrate pilot line has achieved full automation equipment line connection.

The memory sector itself is in a high-prosperity cycle, with companies such as Demingli and Longsys posting sharp earnings growth. Longsys achieved revenue of 38.718 billion yuan in the first three quarters, up 131.37% year on year, and net profit attributable to shareholders of 14.322 billion yuan, up 1909.68% year on year, but third-quarter profit fell sequentially; the company noted that the price increase of memory wafers narrowed and the marginal contribution of price fluctuations to profit is slowing.

A China Merchants Securities research report pointed out that computing power investment in large models continues to climb, and model makers are accelerating the expansion of new revenue sources. As a trillion-dollar-plus consumer market, consumer electronics is inevitably an important direction for model makers seeking new commercialization scenarios.

Robot track industrialization still in early stage

Embodied intelligence is a frontier track combining humanoid robots and large models. This round of third-quarter reports shows that relevant listed companies have initially realized earnings, but the industry as a whole is still in the early industrialization stage of product launches, customer validation and small-batch delivery, generally characterized by sharply rising R&D investment.

Unitree Robotics, a representative A-share humanoid robot complete-machine company, reported in its third-quarter report that revenue in the first three quarters was 1.817 billion yuan, up 55.71% year on year, and net profit attributable to shareholders was 387 million yuan, up 270.54% year on year, but net profit after deducting non-recurring gains and losses fell 18.78% year on year, with single-quarter third-quarter net profit after deducting non-recurring gains and losses down 17.42% year on year.

SWANCOR Advanced Materials said in its third-quarter report that its consumer-grade embodied intelligent robot business made key progress, with Qiyuan Q1 and Qiyuan T1 robots going on sale in September and deliveries starting in October, but no robot business revenue was recognized in this reporting period. The company's R&D investment in the first three quarters was 384 million yuan, up 1182.01% year on year, of which 360 million yuan was invested in robot business R&D.

Core component companies for humanoid robots have already entered the batch delivery stage. Fulin Precision implemented a robot intelligent electric joint project; the company said its robot products have achieved small-batch production and delivery and it is continuing to advance mass production and industrialization. Revenue in the first three quarters was 18.323 billion yuan, up 101.70% year on year, and net profit attributable to shareholders was 645 million yuan, up 98.36% year on year. Orbbec, positioned in 3D visual perception, saw sales revenue from robot vision solutions in the first three quarters reach more than twice the year-earlier level, but the company's overall revenue was basically flat and net profit attributable to shareholders fell 31.62% year on year, because it proactively scaled back its traditional low-margin biometrics business and continued to increase R&D investment in robot vision and embodied intelligence data collection hardware, putting it in a business transition period.

A Huaxin Securities research report believes that deep collaboration between large-model companies and body manufacturers will help accelerate breakthroughs in embodied "brain" technology. As Wang Xingxing, founder of Unitree Robotics, said, with breakthroughs in models such as GPT-6, the tipping point for embodied intelligence is accelerating and may arrive in as little as one or two years.

(Source: China Securities Journal) Middle of the main text Embedded advertisement