Unitree's first post-IPO earnings: 70% revenue growth, R&D spend up 250%
Unitree's first post-IPO earnings report
Unitree Technology's first earnings report as a public company landed on October 28 with a split result. Revenue was 665 million yuan in Q3 2026, which was up 69.93% year over year. Net profit was 112.7 million yuan, down 17.37% year over year, and roughly 50% below what analysts expected.
Why?
Because R&D spending jumped 249.6% in Q3, to 127 million yuan.
Unitree is post-IPO, sitting on 4.2 billion yuan in proceeds, and it is spending aggressively on embodied AI model development, computing infrastructure, and engineering headcount.
What Wang said about the timeline
At the World Physical AI Conference in Hangzhou on October 22, days before earnings dropped, Wang made his most specific public prediction. He said a commercially definitive breakthrough in physical AI would arrive in "1 to 2 years at the earliest, 2 to 3 years at the latest."
That's a significant pull-forward from prior estimates, which ran more toward "several years." It's also the frame within which the R&D surge makes sense: if the critical inflection point is within 2 years, front-loading investment now is rational. If it's further out, the margin compression looks premature.
Wang also said Unitree would expand production capacity to meet potential demand for tens of millions of machines, a statement that's aspirational at current volumes but suggests the internal confidence level is high.
The volume context
Unitree shipped 5,215 humanoid robots in full-year 2025, ranking it first globally by unit volume. The 2026 target is much higher: 10,000 to 20,000 units.
The product line runs from the R1 at $5,900 (launched July 2025, the lowest-cost humanoid on the market) up through the G1 at $13,500, the H2 at $29,900, and the H1 at roughly $90,000.
Humanoid robots now account for 51.78% of Unitree's core business revenue, up from 27.68% in 2024 and 1.88% in 2023. Quadrupeds, where the company shipped over 170,000 Go1 and Go2 units cumulatively, account for another 41.62%.
But where the gross margin on quadrupeds is 56.72% and improving, humanoid gross margin has compressed from 87.67% in 2023 to 63.18% in 2025 as volumes scaled and prices fell. That trend is likely to continue.
Importantly, 73.6% of humanoid revenue comes from research and education customers, with only about 9% from industrial deployment. Unitree is the global market leader in humanoid volume, but it's still mostly selling to labs and universities, not factories.
The IPO and what followed
Unitree listed on Shanghai's STAR Market on August 18 at an IPO price of ¥150.80 per share.
The stock hit an intraday high of +629% on day 1 and closed at +460%. Then it fell roughly 55% from those highs through early October as investors digested the valuation. JPMorgan double-upgraded the stock from Underweight to Overweight around October 20, tripling its price target to ¥995 and citing Wang's accelerated breakthrough timeline.
The stock has since recovered 60 to 70% from its October lows.
As of October 28, Unitree's market capitalization sits at roughly $39.5 billion. The Q3 earnings report, Unitree's first as a public company, was therefore the first real test of whether the IPO price was grounded in anything.
Well, revenue is growing fast and the predicted trajectory is impressive. But profit missed by 50% and the R&D line is now running at 19.1% of revenue, which is high for a hardware manufacturer.
Of course, someone recently said that humanoid robot makers perform like manufacturing companies but are valued like software companies. Making the transition from robots that are interesting/fun/ or for research to robots that customers will lease to do actual work at scale could make that software-style valuation make sense.
The real question
Unitree is the only profitable listed humanoid robot company in the world. That matters. Chinese regulators are actively slowing humanoid robot IPOs amid concerns that valuations are inflated by state-backed project revenues that don't reflect genuine commercial demand.
Unitree's profitability, however compressed, is a structural advantage in that environment.
The nine-month YTD numbers are also notable: 1.817 billion yuan in revenue, up 55.71% year over year, and net profit up 270.54% year over year (mostly a base effect from a large one-time share-based payment charge in the prior-year period, not operational improvement alone).
JPMorgan's revised forecast projects 31,701 humanoid units shipped in 2027 and 111,554 in 2028. If those numbers are anywhere near right, the current R&D investment makes sense. But if the physical AI breakthrough Wang is predicting takes 5 years instead of 2, Unitree will be explaining a prolonged period of margin compression to public market investors who weren't told to expect it.
What Q3 proves is that the revenue is real, the market is growing, and Unitree has made a deliberate choice to invest everything available in winning the AI layer rather than optimizing near-term profit.
That bet could look extraordinarily prescient. Or it could be a problem.
Sources
Written by
I tell stories with data and build startups. I've raised angel and VC funds, led software companies and startups, and chronicled the rise of the mobile economy. Currently I write for Forbes, host several podcasts, and am starting up HumanoidMegaHub. I was TUNE's mobile economist and Singular's VP of Insights, built the research team at VentureBeat, and managed teams creating software for partners like Intel and Disney. In past lives I built websites, e-learning platforms, and mobile apps, and, a long time ago, was named to Folio's top 100 of the media industry's "most innovative entrepreneurs and market shaker-uppers.”