Embodied intelligence startups raise funds every 1-2 months, valuations double between rounds

Fundraising Cycles Accelerate to One or Two Months

The embodied intelligence sector has seen fundraising rounds closing every one to two months over the past two years, with valuations changing month by month. This rapid pace stands in stark contrast to the limited progress in large-scale commercialization and real business orders.

"Startups are basically closing a funding round every one to two months, and the valuation for the next round typically grows by one to two times," said Huang Yi, founder and CEO of RoboParty. He revealed that RoboParty has completed five funding rounds, raising over $100 million in total.

Investors, too, are feeling the acceleration. Liu Ruochuan, Executive Director and Head of Investment Division II at Lenovo Capital, observed that the embodied intelligence sector is clearly overheated from a capital perspective. "Some companies achieve valuations of hundreds of millions of dollars shortly after being founded. For example, world models—whether from new companies, established ones, or even hardware companies—are all talking about world models. Beyond the concept itself being grand, it's also because it's a current hotspot for capital."

Zhang Zhipeng, Associate Professor at Shanghai Jiao Tong University's School of Artificial Intelligence, believes the overall fundraising pace has somewhat moderated. "The rhythm from mid-last year to mid-this year was somewhat abnormal."

Technology iteration is also accelerating. "A proposal made three months ago may already be outdated today," Liu said. Huang Yi agreed, noting that the brightest minds are concentrated in the embodied intelligence field. "Plans can be overturned by the industry in less than three months. Some shareholders constantly ask about product milestones and releases, which puts a lot of pressure on founders."

Commercialization Lags Behind

Against the backdrop of faster fundraising and rapid technological iteration, how is commercialization progressing?

Liu acknowledged that the industry's business闭环 and commercialization are relatively lagging. "This requires efforts across the entire industrial chain—not just tech companies and entrepreneurs in the main赛道, but also upstream and downstream ecosystems, as well as capital."

"Looking at past industry cycles, it's very normal to see several ups and downs," Liu added. "The market needs to have reasonable expectations."

Zhang Zhipeng pointed out that neither the hardware nor the application scenarios for embodied intelligence are ready. "For hardware, it's hard to find a platform that can operate stably over the long term. For scenarios—such as delivery, micro-fulfillment centers, or homes—it's still unclear which one can achieve large-scale stable deployment within a few years. We're still in the exploration, trial, and competition phase."

He believes that the number of companies meeting expectations is not that large. "There are about 400 embodied intelligence companies, but far fewer than expected are truly committed to deep tech development. Many companies focus more on marketing around trending keywords—from VLA (Vision-Language-Action models) to world models, and now to RSI (Recursive Self-Improvement)—without actually digging deep into the technology."

How Should Founders Balance Product and Ecosystem Investment?

When fundraising节奏 and commercialization are misaligned, how should founders manage their investment in products and ecosystems?

Huang Yi's answer: calculate how many years the cash on hand can sustain operations. "There's no need to be like some embodied intelligence 'brain' companies that raised three to four billion but only spent one to two hundred million. That's a waste of resources and doesn't advance the industry."

In his view, a key quality for a founder is restraint—knowing what to do and what not to do. "That's what makes the best founder." He plans to focus his energy over the next three years on "selling robots, selling hardware." "A company's commercialization path must generate its own cash flow. Just because you've raised a lot of money doesn't mean you can spend recklessly on models. That's irrational."

Can Capital Build a Moat?

Capital is also changing the rules of the game. Currently, fundraising leverage is concentrating among "head" companies—but "head" may refer to valuation leadership rather than technological or product leadership. Some investors believe that valuation leaders have raised enough money, have strong cash reserves, and are safe—at least from going under.

Liu Ruochuan cautioned: "The market should not underestimate the power of capital, but neither should it mythologize it. We need to view it objectively and rationally. Whether capital alone can build a moat—the mobile internet era over the past decade has given us a clear answer. There are classic cases where capital burned through money to build platform value, and others where heavy capital bets ended in nothing."

"Applied to the hard-tech sector, we've seen capital concentration over the past seven to eight years. Frankly, when evaluating projects, fundraising ability may be one dimension to assess, but it is by no means the only one," Liu said.

Qi Lin, Executive Director at Lochpine Capital, offered another perspective: "Technology is still advancing, and the embodied model赛道 is iterating continuously. Capital is also driving the market forward. The two are mutually reinforcing—neither dominates the other."