by ZHOU Mo, JIN Jing

China's latest government work report places technological innovation – particularly hard tech – at the center of economic policy, prompting investors to seek the next trillion-yuan growth sectors.

At the opening of China's annual "two sessions" meetings on March 5, Premier Li Qiang said China would accelerate new growth drivers by nurturing emerging and future industries while promoting new forms of the "intelligent economy".

Investors interviewed by Jiemian News said the policy signals strengthen expectations that several frontier sectors could become major investment themes in the coming decade.

Embodied intelligence emerges as a key focus

The report calls for developing new pillar industries such as integrated circuits, aerospace, biomedicine and the low-altitude economy, while fostering future industries including future energy, quantum technology, embodied intelligence, brain-computer interfaces and 6G. Several of these concepts – including "new pillar industries" – appeared in the report for the first time.

DING Hao, a partner at Beyond Capital, said the shift in language from "emerging industries" to "new pillar industries" suggests some sectors are approaching large-scale commercialization. The report also encourages state-owned enterprises to open application scenarios, which he said could accelerate the commercial validation of new technologies.

Early examples include low-altitude logistics networks in parts of the Greater Bay Area.

The report pledged to make better use of the national venture capital guidance fund and encourage government-backed funds to act as "patient capital".

"This is a big boost for those of us investing in early-stage deep tech," said LIU Gang, a partner at Alpha Startup Fund.

Embodied intelligence – often linked to humanoid robots and AI-powered systems – is drawing strong interest from venture investors. Several said the sector could eventually follow a trajectory similar to the early development of China's electric vehicle industry, although both the technology and its commercialization path remain at an early stage.

"Complete robot systems are where the real breakthroughs will happen," one investor said.

In the longer term, Liu said three factors would be key: tight cost control, an "iOS-level" operating system and more powerful world models, which together would lower barriers for consumers, build an ecosystem and strengthen the underlying technology.

Ding added that his firm is also investing in frontier technologies including 6G, quantum computing and nuclear fusion. 6G could be deeply integrated with artificial intelligence, he said, while quantum computing – though still largely at the basic research stage – has vast long-term potential.

The government work report said China will build international innovation centers in Beijing (the Beijing-Tianjin-Hebei region), Shanghai (the Yangtze River Delta) and the Guangdong-Hong Kong-Macao Greater Bay Area to better link technological and industrial innovation.

Waiting for a "killer app"

The government work report, for the first time, introduced the concept of a "new form of intelligent economy", calling for wider adoption of intelligent terminals and AI agents and faster commercialization of AI across key industries.

Earlier this year, Chinese foundation-model developers Zhipu AI and MiniMax – both valued at tens of billions of yuan privately – went public in Hong Kong. Their share prices surged soon after listing, briefly pushing their market capitalization above HK$300 billion.

"This is the first time global secondary-market capital has priced AI foundation models," said WANG Jie, co-director of a Shenzhen-based AI research institute, adding the technology had moved beyond its "childhood" stage.

Strong public-market performance by GPU and large-model companies could lift private valuations and shift investor attention to applications, Liu Gang said.

Kai-Fu Lee, chief executive of 01.AI and chairman of Sinovation Ventures, said large-model technologies are evolving rapidly and generative AI – which he calls "AI 2.0" – is moving toward large-scale commercial deployment.

Yet investors say China still lacks a consumer-facing AI product with mass-market penetration – a true "killer app". Wang said one could emerge as early as 2026 as technological capabilities expand and inference costs continue to fall, declining by roughly 90% every 12 months.

The initial wave of foundation-model development in 2023 has largely passed, another investor said, with attention now shifting to applications, though monetization remains uncertain.

"Everyone is racing to win users, but few have figured out how to make money," the investor said, pointing to ByteDance's AI assistant Doubao as an example.

Ding said most consumers currently use AI mainly for information searches and learning support, while tolerance for errors in financial or decision-making scenarios remains low. Near-term opportunities may therefore lie in B2C applications with higher tolerance for mistakes, particularly in content creation and marketing.

The enterprise market offers clearer paths to profitability. Wang said the biggest opportunities lie in "AI productivity" and "digital employees", with AI coding emerging as a key growth segment.

Data from Ding's portfolio companies suggest AI coding systems can already boost junior engineers' productivity three to five times, though they cannot yet handle complex system architecture tasks independently.

Building the AI infrastructure

The government work report also signaled stronger support for open-source AI ecosystems.

Wang said open source originally emerged as a bottom-up strategy pursued by Chinese developers to narrow the gap with global leaders. Its inclusion in the report signals it is now a national priority.

Open-source models from Alibaba's Qwen, Moonshot AI, Zhipu AI and MiniMax are gaining traction among global developers. According to OpenRouter data, Chinese AI models surpassed US models in total usage for the first time in February 2026, occupying four of the top five positions globally.

LI Lecheng, China's industry minister, said during the two sessions that open-source models released by Chinese companies last year ranked first globally in downloads, lowering barriers and costs for AI use.

Ding said open-source models are key to scaling the AI industry and could eventually dominate general-purpose models, particularly as embodied intelligence expands into industries where many smaller customers cannot afford customized systems.

According to the Ministry of Industry and Information Technology, China's computing infrastructure continues to expand rapidly. The country's three major telecom operators provided about 938,000 data-center racks for external services, while schedulable intelligent computing capacity exceeded 94.4 exaflops.

Ding said state involvement in computing infrastructure is inevitable as demand for chips and computing power surges and most start-ups cannot afford their own computing centers.

"In a sense, computing power is the energy of the future," he said.