This spring, investment promotion teams became regular visitors to financial advisor He Ziyang’s office. She works at an advisory firm serving hard tech companies, where she connects embodied intelligence startups with investors. On her busiest day, she received four government delegations, from Tianjin, Shenzhen, Chengdu, and Xiamen. Most represented district-level investment promotion centers. Others came from prefecture-level cities.
In general, hard tech refers to technologies built on substantial scientific or engineering advances, such as robotics and semiconductors.
Some teams were candid about why they had entered the sector: they were following policy without a clear strategy of their own.
But some had plans that seemed unrealistic. One county-level city with only a few hundred thousand residents had a strong manufacturing base. Its team could map the local industrial network in detail: ready-to-use factories, comprehensive support services, and plenty of small and midsize suppliers and customers. All it lacked was an anchor company.
He Ziyang sensed their uncertainty. Many seemed compelled to pursue a sector they did not fully understand.
State capital was less visible during earlier investment booms in consumer goods, entertainment, and the internet. In hard tech, commercial funds alone struggle to meet broader policy objectives, giving state-backed investors a larger role. Their presence is especially prominent in embodied intelligence, one of the most closely watched hard tech sectors of the past two years.
ITjuzi data show that China’s embodied intelligence industry recorded more than 300 financing deals in the first half of 2026, up 137% year-on-year, with disclosed funding totaling RMB 93.5 billion (USD 14 billion). State-backed institutions participated in 42% of large financing deals. Local state investors, corporate investors, and leading venture capital firms now commonly invest together.
By late 2025, state-backed investors were entering in force. Shanghai’s embodied intelligence fund and Shenzhen’s group of robotics funds became important sources of new capital. Beyond the biggest cities, robotics funds in Hangzhou, Suzhou, Hefei, Wuxi, Ningbo, and Nanjing also joined the competition.
“When the big local funds were set up last year, we thought this year would certainly be strong. And it has been. Leading companies are getting RMB 1 billion (USD 150.1 million) or several billion in a single round,” Liu Kaijie, an investor at an established hard tech fund, told 36Kr.
For a while, investing in embodied intelligence looked like an open-book exam with no wrong answer.
“Everyone is investing in a prescribed brief. You either invest in Musk or in ‘Marx.’ There is greater certainty in aligning with the overall direction. That is the consensus,” an investor at a state-backed fund said. Several other interviewees expressed the same thought in different terms: follow a technology leader, or follow the state’s priorities.
Hefei’s model of government-led industrial investment had shown how a relatively small commitment could produce a much larger payoff. As investors’ attention moved from complete robots to upstream components and then world models, state capital followed.
The mood changed sharply in September. Unitree Robotics lost half its market value after listing, and the founder of Mech-Mind Robotics publicly criticized industry practices on WeChat. Earlier in the year, Hong Kong had also adjusted its Chapter 18C listing rules for specialist technology companies.
IPO prospects for second- and third-tier companies were fading. Several investors told 36Kr that financing pressure would become acute after November this year. Many companies still had cash from previous rounds, but new funding would come at lower valuations, while their investors faced fewer viable exits. That threatens both the financial returns local state investors expect and the industrial growth they hope to secure.
“Embodied intelligence changes too quickly. There is a small iteration every three months, and six months can overturn every previous judgment. The market is deeply uncertain,” an investor at a district-level institution in Shenzhen said.
According to 36Kr, Unitree ranked second globally in humanoid robot shipments, with more than 5,000 units shipped in the first half of 2026. Although its market value remains roughly in line with its pre-IPO valuation, some investors saw its sharp post-listing decline as a sign that the industry should slow down.
Paying to attract a robotics industry
What does embodied intelligence actually bring to a locality?
One state investor repeatedly referred to an ancient story about paying a fortune for the bones of a fine horse. During China’s Warring States period, a ruler struggled to find an exceptional horse. A minister bought the bones of a dead one at a high price to demonstrate the ruler’s determination: if he would pay for a dead horse, imagine what he would offer for a living one. Within a year, the ruler obtained three fine horses.
More than two millennia later, local governments are using a similar signal to attract businesses: a conspicuous investment can demonstrate their commitment to an industry.
Cities seeking a place in the push for “new quality productive forces” and the 15th five-year plan have limited options. Commercial spaceflight, quantum technology, and controlled nuclear fusion depend heavily on national-scale research facilities, leading scientists, and specialized industrial infrastructure. Few cities can accommodate them.
Embodied intelligence offers more flexibility. Companies can base robot assembly, component manufacturing, software development, and data operations in different locations. A city does not necessarily need a manufacturer of complete humanoid robots. A company making joint motors or reduction gears can help meet near-term investment targets, offer the prospect of future tax revenue and industrial growth, and give officials a basis for ambitious local robotics plans.
Districts may be assigned different roles to prevent destructive competition, according to 36Kr. One may focus on key components, another on robot manufacturers and artificial intelligence developers, and a third on makers of dexterous robotic hands.
Some companies have adjusted their fundraising accordingly. They study the industrial resources available alongside state funding, divide their operations to match local needs, and place headquarters, R&D centers, production lines, and data centers in different cities.
“If a locality lacks complete machines, they pitch the robot itself. If it wants components, they emphasize motors and reduction gears. If it wants a physical presence, they build a data center,” one investor said.
For some local state investors, the priority is to document that they have met their investment promotion targets. Companies can provide that evidence relatively cheaply.
“What they fear most is not necessarily that a project performs poorly for a while. Hard tech allows for failure, and policy keeps emphasizing exemption from liability when duties have been properly performed. What they fear is a failed project with nothing in the files to show for it,” the investor said.
Companies adept at this process first study gaps in a city’s efforts to attract investment. They sign agreements to establish a base, secure factory premises, buy equipment, arrange some orders with industrial partners, issue press releases on schedule, and display photographs of officials visiting.
“The paperwork has to tell a complete story first,” the investor said.
Whether the company creates real value can be studied over time, the investor said. “In the end, a company whose commercial prospects are barely formed is already described in official documents as a key enterprise in the local robotics supply chain.”
“Everyone thinks they have secured the most important piece. In the end, they are all carrying the sedan chair together, and nobody wants to be the first to ask whether anyone is actually sitting inside.”
State capital follows different mandates
China’s state investment system spans regions with different priorities and overlapping lines of authority. Central and local responsibilities intersect, as do those of different local governments. Officials answer both to departmental hierarchies and to the jurisdictions in which they work. State capital is therefore far from a single pool of money. Funding sources, performance measures, risk appetite, and investment styles vary.
One category of state fund serves a parent group’s industrial needs. Examples include investment platforms under centrally administered state-owned enterprises, such as CITIC-affiliated funds and CRRC Capital. These institutions have substantial resources, including earmarked funding and transfers to support industrial development. They are judged less on the return from an individual investment than on whether it fills gaps in the group’s supply chain or opens new business opportunities.
CRRC’s transformation and upgrading fund, which invested in AI² Robotics, described the attraction as a “natural fit between their ecosystems.” CRRC has advanced manufacturing operations and wants to know whether embodied intelligence can be integrated into its production lines and businesses.
The second category serves local targets for attracting investment. District- and county-level state platforms and industrial funds have been among the most active investors in this boom. Their obligation to preserve and increase the value of state assets tends to make them cautious. They often prefer more mature, mid- to late-stage companies with ten-figure RMB valuations, closer to the private equity stage. The central measure is what the company will bring to the local economy.
“We care more about whether a company can put its R&D center, production lines, and supporting teams here, and whether it can bring upstream and downstream component suppliers and data service companies together into a cluster. Those are the fund’s first priorities. The company’s short-term valuation and when it will list come second,” Wu Jianyu, who heads a district-level investment promotion center in a second-tier city, told 36Kr.
The third category consists of commercially oriented state funds whose primary goal is financial return. Shenzhen and Shanghai have piloted reforms that classify funds by their objectives, allowing some to be assessed on investment performance rather than their success in attracting local businesses. These funds do not require companies to build local factories. They relax local investment requirements and place greater emphasis on growth potential and returns when they exit.
Examples include TopoScend Capital under Shenzhen Investment Holdings, an AI industry fund managed by SCGC, and some subfunds under Shanghai State-owned Capital Investment (SSCI). Such institutions have appeared among the backers of leading embodied intelligence companies including Astribot, Agibot, and X Square Robot.
Regional differences further shape investment decisions, as local officials compete for both government recognition and market opportunities.
Competition is particularly intense in the Yangtze River Delta, where some localities pay little attention to valuation. County-level cities with established manufacturing bases have assembled suppliers across the robotics chain and are pursuing new companies aggressively.
“If the team is star-studded and the story is big enough, they will put in RMB 100–200 million (USD 15–30 million) to bring it over, even at an asking valuation of RMB 6 billion (USD 900.8 million),” He said. Their priority is to secure a prominent company. The price comes later.
“The city you are in largely determines whether you think the market is hot or cold,” said Chen Peng, an investor who has followed hard tech for years.
His institution favors founders who were senior executives at major technology companies and has backed several embodied intelligence businesses. In the second half of this year, however, its investment committee postponed a deal he proposed. A startup led by a senior technical executive from a leading company would previously have been highly likely to win approval.
“Shenzhen may be tightening investment approvals today, while industrial funds in Guangxi and some counties in northern Jiangsu are still expanding. The pace of tightening and easing is highly regional. As the window shifts, the same city can reach completely opposite valuations for the same type of project,” Chen said.
Neither approach is necessarily better. A fundraising executive at an embodied intelligence company said some regions offered more attractive incentives on paper, then delivered less than promised after the company moved in. Places with longer processes and detailed, explicit rules could be slower initially but more consistent in execution and policy. The choice depends on whether a company needs speed or certainty more.
Competing for companies with bargaining power
The regions with the most capital do not always secure the best companies. Access to resources, investment terms, and the ability to pursue deals also matter. Brokerage-affiliated funds, known for aggressive dealmaking and strong incentives, can cover as many opportunities with a few dozen staff as large funds in major cities employing several hundred.
Still, the biggest cities offer advantages. Concentrations of robotics and AI startups, supported by local industrial networks, give investment teams better information for sourcing deals and conducting due diligence. In Shenzhen’s Nanshan district, supply chains, talent, customers, and policy support can all be reached within a half-hour drive.
But those advantages are only a starting point. Leading embodied intelligence companies have gained substantial negotiating power, making traditional offers such as talent subsidies, tax breaks, and industrial land less compelling. Wu said leading companies were reluctant to accept rigid local investment requirements.
“If a locality wants a company to establish operations there, the investment threshold is often a nine-figure RMB sum. It takes a major commitment. Many teams also criticize the long decision-making chains inside state investors. Conversely, some institutions make the absence of local establishment and local investment conditions their main advantage in attracting companies,” he said.
Provincial and district-level funds are increasingly taking different approaches. Most provincial investment companies no longer insist that businesses establish operations or register locally. Since the beginning of this year, many have shifted toward commercial investments, dropping local requirements that can make it harder to win the best deals. District and county funds still place more weight on securing a local presence.
This shift also reflects central efforts to curb excessive competition among local governments for investment. That competition created risks and increased pressure on local fund managers to demonstrate that they had performed their duties properly, making them more cautious.
Economic development is both a central task and a measure of official performance. It affects local leaders’ standing and promotion prospects, as well as performance assessments for lower-level civil servants. Officials invest to build political capital, while remaining subject to party discipline and the law. Scholar Lan Xiaohuan describes this as a “carrot-and-stick” incentive system: officials must pursue growth while managing the personal consequences of failure.
Letting commercial investors set the price
State investors have generally kept their distance from the most exuberant valuations.
“We have seen companies with low eight-figure RMB revenue valued at more than RMB 10 billion (USD 1.5 billion). We simply cannot invest in something like that,” one investor said.
State funds rarely push valuations that high themselves. Liu said most let commercially driven investors establish a valuation before following. After securing prospective allocations from three or four local investors, a founder may hear: “Find a commercial investment firm to set the price. Once it has done that, we can sign.”
Letting commercial investors set the price gives state funds a basis for defending their decisions. What looks cautious is also a calculated way to reduce exposure to criticism: these investors seek certainty more than exceptional returns.
“If any company valued above RMB 10 billion says it is opening a round, everyone will compete for an allocation because the certainty is high,” said an investment manager at an industrial investment platform under a central state-owned enterprise. “Steady returns matter more than large returns.”
Industrial promises face closer scrutiny
Even companies valued at RMB 10 billion may fail to deliver that certainty. Local funds are becoming more cautious. They have not abandoned industrial development in favor of financial returns, but they are looking more closely at whether investments deliver a functioning local business.
Previously, signing an agreement to establish a base, installing a production line, and buying a batch of robots could be enough to meet an investment-attraction target. Since the State Council issued guidance on risks, regulation, and the development of private investment funds, officials approving investments must explain why the money was committed and whether the promised operations are running.
Investors are also reassessing their exit options. Chapter 18C remains available, but 36Kr reports greater demands for commercial progress and credible technology. High valuations and prominent shareholders alone are less likely to secure a listing.
Every investor and portfolio company faces the same question: without an IPO, how will investors get their money back? One optimistic view held that a large cash reserve would keep a company alive even if commercialization stalled in the short term. But survival alone does not give investors a way to recover their money.
The CEO of a leading embodied intelligence company offered a hypothetical example: a business that raises RMB 5 billion (USD 750.6 million) and loses RMB 300–500 million (USD 45.0–75.1 million) annually might have enough cash for nearly a decade. But a prolonged lack of commercial progress creates mounting pressure to buy back investors’ shares. Disappointing revenue and product deployment also weaken its competitive position and make it harder to retain key employees.
State investors could then lose on both fronts: equity bought at a high valuation would be difficult to sell, while the local operations and supply-chain development expected from the investment might fail to materialize.
Even an IPO does not end the test.
“Capital carries the company into the sedan chair. The IPO lifts the curtain so everyone can see what is actually inside. Before, a dozen or so institutions could sit behind closed doors and believe in the future together. After listing, tens of thousands of shareholders will be watching every day and asking about revenue, profit, and the cash flow,” the CEO said.
Securing funding and reaching an IPO leave the harder task of sustaining the business. A relatively favorable outcome would be for the company to remain stable long enough for investors to sell shares in stages after lockups expire, easing pressure to produce returns.
Ultimately, though, the business has to make money. For state investors, protecting their reputations takes priority over cashing out, encouraging them to seek less risky entry points.
Funding shifts to early-stage and pre-IPO companies
State investment is increasingly concentrated at the earliest stages of a company’s development and just before an IPO. At the early stage, investors select companies, help set valuations, and work closely with them to capture returns from subsequent growth. This is the policy of “investing early, investing small, and investing in hard tech.” At the other end, groups of institutions back prominent pre-IPO companies, seeking to protect capital and improve the likelihood of an exit through listing.
Pre-IPO valuations are already close to public-market levels. Investors place more weight on the prospect of a listing than on the underlying business, viewing an IPO as some protection for their investment. That perceived security usually comes at a substantial premium. The cost of entering late is pushing both state investors and USD-denominated funds to seek opportunities within a startup’s first three funding rounds.
“People used to think state capital was always the last buyer in. Now we see that is not the case. They are investing in the first three rounds too,” Chen said.
Founders often have a different preference. Many view state capital as more suitable later, when it can support a listing and improve the mix of shareholders.
“Taking state capital early comes with requirements around local production capacity, tax contributions, and other commitments. At an early stage, we would rather not be constrained by those rigid conditions,” the founder of an embodied intelligence company said.
“We want to invest earlier too, but founders all think state capital should be the last buyer in,” one investor said.
Companies between these stages face the greatest concentration of risk. They rely on customized orders for growth and successive funding rounds for cash, while their products have yet to generate steady repeat purchases. Leading companies are absorbing the limited available capital, leaving those in the middle struggling to raise money.
The limits of investing by policy
Whatever happens next in embodied intelligence, following policy remains a dominant investment strategy.
“It is not that venture capital has suddenly developed long-term patience. It is that policy support allows a project’s failure to meet expectations to be attributed to uncertainty in an exploratory stage of the sector,” one investor said.
Investing alongside others gives state fund managers a measure of protection from criticism. Participation by a leading venture capital firm can be presented as a collective judgment reflecting market consensus. Local operations provide documented evidence of industrial development even when commercialization is slower than expected. Investors who independently back an unfamiliar sector face greater personal pressure if it fails.
“Everyone looks upward: at policy, at the official line, at what different levels of government have been talking about lately, at which industrial park needs projects, and at which sector is easiest to put in a report. People are not discovering the future. They are guessing the future from policy documents,” the investor said.
That may remain the norm for years. Funding for embodied intelligence has tightened for now, but hard tech remains a development priority. Investors still need to find companies across the supply chain with sustainable businesses.
The government cannot provide unlimited support to emerging industries. “A strong government is defined not only by its ability and resources to support companies, but also by its ability to refuse them help,” Lan wrote.
Government intervention was part of East Asia’s economic rise. State support helped China’s display-panel and solar industries become globally competitive, while South Korea relied on public investment to develop shipbuilding and electronics industries it had not previously possessed. Those industries matured by eventually operating without that support.
“New energy and semiconductors have followed similar paths. Now it is robotics’ turn. We have to respect market sentiment,” the investor said.
Liu Kaijie, He Ziyang, Wu Jianyu, and Chen Peng are pseudonyms used at the interviewees’ request.
KrASIA features translated and adapted content that was originally published by 36Kr. This article was written by Huang Nan for 36Kr.
Note: RMB figures are converted to USD at rates of RMB 6.66 = USD 1 based on estimates as of November 2, 2026, unless otherwise stated. USD conversions are approximate and, where appropriate, rounded for ease of reference. They may not fully match prevailing exchange rates.