China’s primary market is showing a striking contradiction in 2026. Venture investors are still talking about how hard it is to raise funds and how difficult it is to exit investments, yet a handful of hot sectors — large AI models, embodied intelligence, fusion, and quantum computing — are once again seeing valuations jump and financing rounds get chased by capital.
Incomplete statistics cited from Yicai show that China added 19 embodied intelligence companies valued above RMB 10 billion in the first half of 2026 alone. Among them, Galaxy General, Zibianliang, Zhipingfang, and Xinghaitu have already exceeded RMB 20 billion in valuation. Some companies moved from the RMB 10 billion level to the RMB 20 billion range in just six months.
Valuations in large-model AI have climbed even faster. Moonshot AI completed a Series F financing of more than $3.5 billion in July this year, bringing its post-money valuation to $35 billion. It then launched a pre-IPO round, with the market-disclosed target pre-money valuation rising to $50 billion, or about RMB 340 billion. At the start of the year, its valuation was still around $10 billion.
The core point in the article is that money has not started flowing back evenly across the entire private market. What is happening instead is that limited active capital is being pushed harder into a small number of projects that look more “certain.”
Exits, not projects, remain the biggest constraint
To understand this latest round of valuation increases, the article says the first step is to understand what has hurt the private market most over the past few years. The main problem is not a total lack of good technology, and it is not a complete lack of funding. It is exits.
That is the biggest difference between primary and secondary markets. Investors in private companies cannot sell whenever they want. Returns usually depend on an initial public offering, a merger and acquisition deal, or a transfer of old shares. If the exit path is unclear, even a high paper valuation is hard to convert into cash returns for a fund.
That is why investor interest can change quickly when a company suddenly has a clearer IPO trajectory. The article presents Moonshot AI as a typical case. Public reports show the company completed multiple rounds of financing in the first half of 2026, with its valuation rising from about $10 billion to $35 billion, before launching a pre-IPO financing at a disclosed target pre-money valuation of $50 billion. The market also saw demand for the round clearly exceed the original plan, leading to an early close.
What capital is buying here is no longer only the next decade of AI growth. It is also buying something else: a shorter distance to exit. If a company is still several years away from a possible IPO, investors must bear a long period of uncertainty. If it has already started a shareholding reform, listing guidance, or pre-IPO financing, some funds may be more willing to accept a higher valuation.
That creates a choice that looks contradictory on the surface: the price is higher, yet some institutions believe the risk may actually be lower.
Listed peers are repricing private companies
The article says a second source of “certainty” is coming from the secondary market.
In the past, valuations for AI and robotics companies in the primary market were often built around revenue forecasts, technical teams, market size, and peer financing rounds. Now, listed companies in the same sectors are becoming a more important reference point. When a public company in a comparable field achieves a high market value, private companies in that field gain a new valuation anchor almost immediately.
That is one reason financing enthusiasm in embodied intelligence picked up so clearly in 2026. As a number of leading companies push forward capital-market plans, and as firms such as Unitree draw heavy attention from the market, investors naturally start recalculating the possible value of comparable companies that are still private.
The logic is simple. If an unlisted robotics company is valued at RMB 10 billion and the market believes a listed peer could reach RMB 30 billion, then RMB 10 billion may still look like it leaves room for further upside.
But that is also where the risk sits. The secondary market can reprice assets every day. The primary market cannot adjust so easily. If a listed peer falls from RMB 30 billion back to RMB 15 billion, the “safety cushion” for a private company valued at RMB 10 billion can disappear very quickly.
In that sense, the public market can push private valuations higher, and it can also pull them lower. As the number of unicorns rises, the private market is becoming more dependent on public-market sentiment.
The next round getting pricier is becoming another form of certainty
Beyond IPO expectations and listed-peer benchmarks, the article argues that the transaction process itself is beginning to create certainty.
It gives a simple example. Suppose a company was valued at RMB 5 billion in the last round. In the next round, it may raise only RMB 500 million and sell a small amount of equity, yet still push the headline valuation to RMB 10 billion. Existing shareholders then record paper gains. New investors accept the RMB 10 billion price because they believe the following round may happen at RMB 15 billion or even RMB 20 billion.
As long as new money keeps coming in, that valuation system can continue to operate. Under that setup, the investment question begins to change. The old question was: what will this company be worth in five years? For some investors now, the first question becomes: will someone come in at a higher price in the next round?
The article says those two approaches may look similar, but they are fundamentally different. The first depends on the company eventually generating cash flow and profit. The second depends on whether the transaction chain can continue. A rising valuation by itself does not prove that underlying value has increased.
A RMB 10.6 billion fusion valuation points to a scarcity premium
This concentration of capital has spread from AI and robotics into earlier-stage hard-tech sectors.
In July, ENN Fusion completed its first external financing. After the pre-A round, its post-money valuation reached RMB 10.6 billion. The funds will be used for construction of a third-generation spherical ring proton-boron fusion device, technology iteration, and expansion of its research and development team.
The article notes that this remains a field still far from large-scale commercialization. Even so, since the start of 2026, at least 12 private fusion companies in China have announced new financings. Since 2022, these companies have disclosed about 24 financing events, with 14 of them concentrated in the first seven months of this year.
Why are investors willing to assign valuations above RMB 10 billion before the business model is fully proven? The explanation offered in the article is not only technical promise, but also scarcity. When policy signals, industrial trends, and capital consensus all converge on one sector, and the number of companies with mature teams and technical accumulation is limited, investors are willing to pay a scarcity premium.
Still, scarcity does not equal eventual success. Fusion, quantum computing, and general-purpose robotics may all carry trillion-yuan potential in theory, but they may also need years, or even more than a decade, before real commercial validation arrives.
Risk is shifting from technology to valuation
That also explains why this unicorn wave is different from the previous cycle of internet investing, according to the article. Capital does not know which embodied-intelligence route will definitely win, and it does not know which large-model company will ultimately build a stable business model.
The technology uncertainty has not gone away. What investors are doing instead is looking for other forms of certainty that can be quantified: whether a company has an IPO timetable, whether top-tier institutions are already in the cap table, whether the next financing round is already lining up, and how much listed peers may be worth.
As a result, the nature of risk has shifted. Technology risk has not disappeared. It has been temporarily covered by rising valuations.
The article says the higher the valuation climbs, the more a company has to prove. At RMB 5 billion, a company may only need to show that its technology has potential. At RMB 20 billion, it may need to show that customers are willing to pay. At RMB 50 billion, it may need to prove revenue can keep expanding. Once the valuation reaches the RMB 100 billion range, the market will eventually ask about profit and cash flow.
The earlier a valuation pulls forward future expectations, the more operating performance the company will need to deliver later.
The real question is whether valuations can hold
The article ends by arguing that the return of large numbers of RMB 10 billion unicorns in 2026 does not automatically mean a bubble, and it should not be read simply as a fresh bout of capital frenzy. AI, robotics, fusion, and quantum technology do represent major future technology directions, and some companies in those sectors have already shown real orders, revenue, and technical progress.
The key distinction is whether capital is buying long-term corporate value, or a transaction opportunity created by a temporarily open exit window.
If a company can turn each financing round into better technology, larger revenue, lower costs, and stronger commercialization ability, then a higher valuation has an industrial basis. If the main logic is instead that someone invested in the last round, the next round will be more expensive, and the company is preparing for IPO after that, then the so-called “certainty” is really just that the transaction chain has not broken yet.
The article points to previous phases in autonomous driving, computer vision, and the last semiconductor investment boom as examples of similar patterns. When the market is hottest, institutions scramble for allocations. Once financing conditions change, investors eventually come back to the same question: what did all of that money actually buy?
That may be the most useful lens for reading the surge in unicorns in 2026. The private market has not suddenly become more willing to take risk. The opposite may be closer to the truth. As technology and business models become harder to predict, capital is searching more aggressively for certainty.
What is being chased now, however, is no longer certainty that a company will definitely make money in the future. It is certainty that a company has a higher chance of listing, and a higher chance that someone will pay more in the next round.
Those two forms of certainty may shorten the waiting period for investors, but they cannot replace a company’s ability to create real value. That is why the more important question behind the addition of 19 embodied-intelligence unicorns above RMB 10 billion in just half a year is not how many such companies China suddenly has. It is how many of today’s RMB 10 billion, RMB 20 billion, or even higher valuations will still stand when capital starts asking again for proof in revenue, profit, and cash flow.
The article cites Yicai, Shanghai Securities News, Securities Times, Caijing Magazine, and public financing information from companies. It also carries a risk disclosure stating that the analysis is based on public financing and market materials and does not constitute investment advice. Valuations of unlisted companies mostly come from private fundraising transactions, have limited liquidity, and do not equal realizable enterprise value. Some financing and listing plans are based on market disclosures, and the final situation should be subject to formal announcements by the companies and regulators.
The original article was published by the WeChat public account BT Finance, authored by Jiang Xu.