Gasgoo Munich- On the evening of October 28, Unitree released its first earnings report since going public — covering the third quarter of 2026.
The headline numbers look impressive: net profit attributable to shareholders for the first three quarters reached 387 million yuan — a 270.54% surge from a year earlier.
Yet on the same page, recurring net profit — stripping out one-off items — stood at 349 million yuan, an 18.78% decline year-over-year.
That divergence reveals the reality of Unitree’s operations: revenue is climbing, but profitability isn't keeping pace.
Image Source: Unitree
Revenue Accelerates, While Profit Nearly Halves Quarter-Over-Quarter
For the first three quarters of the year, Unitree generated 1.817 billion yuan in revenue, a 55.71% jump. The third quarter alone contributed 665 million yuan — a 69.93% surge that outpaced the overall nine-month growth rate by 14 percentage points.
Driving that growth is the G1, Unitree’s compact humanoid robot. Since its launch in May 2024, cumulative production has reached roughly 10,000 to 20,000 units.
Unitree’s profits, however, are moving in the opposite direction.
In the third quarter, attributable net profit hit 113 million yuan — a 17.37% year-on-year drop. That represents a near-halving from the 224 million yuan recorded in the second quarter, dragging the single-quarter net margin down to around 17%.
A closer look shows the nine-month profit surge relied on two factors: revenue growth boosting gross profit, and a low base from last year. Specifically, the company recognized a one-time share-based payment expense of about 349 million yuan in the same period last year, which was booked as a non-recurring item and severely depressed the prior year’s baseline.
In other words, a significant chunk of that 270% gain was handed over by last year’s accounting adjustments, rather than earned through operational strength this year. Strip out those one-time factors like share payments, and the 18.78% decline in recurring profit offers a much clearer picture of the underlying business.
So, what is eroding Unitree’s profits? The expense statement tells the story.
First, there’s R&D. Research spending totaled 263 million yuan in the first nine months — a 191.48% spike that claimed 14.47% of revenue. By the third quarter, that ratio had climbed to 19.10%.
According to Unitree, the bulk of that funding is flowing into embodied large models, computing clusters, and R&D personnel. In September, the company officially released WLA-1.0, a 6-billion-parameter model, and open-sourced it. The model was reportedly trained on roughly 2,500 hours of real-world data.
Then there are sales costs. Selling expenses reached 230 million yuan over the nine months, soaring 202% and pushing the expense ratio from 6.51% to 12.65%. That spending reflects a brand push during the Spring Festival Gala in the first half, alongside an expanded sales force.
Beyond rising expenses, shrinking gross margins are also squeezing profitability. Unitree’s gross margin fell by about 5.7 percentage points in the first three quarters, which — scaled against revenue — translates to roughly a 100 million yuan hit to gross profit.
Still, the IPO has significantly bolstered Unitree’s balance sheet.
By the end of September, the company held 7.354 billion yuan in cash, with total assets reaching 10.156 billion yuan. Inventory surged 177% to 1.019 billion yuan, signaling that Unitree is stockpiling components to meet upcoming orders.
Image Source: Screenshot from Eastmoney
70% of Humanoid Revenue Relies on Education and Research; Changing the Mix is Key
If rising expenses are a short-term strategic choice, the revenue mix poses a longer-term challenge.
The prospectus reveals that in the first three quarters of 2025, research and education accounted for 73.6% of Unitree’s humanoid robot revenue. Commercial use — such as shopping malls and exhibitions — made up 17.39%, while actual industrial applications contributed just 9.01%. Notably, the top five customers represented only 12.09% of total revenue.
Unitree is acutely aware of this imbalance and is actively working to fix it.
Recently, the company launched the "Yu Chuang Plan," aiming to incubate more than 200 high-quality ecosystem partners over the next five years. The estimated collaboration scale could reach billions of yuan, channeling industry demand toward partners in areas ranging from urban governance and emergency response to industrial and commercial services.
Meanwhile, Unitree founder, chairman, and CEO Wang Xingxing outlined the company’s strategic roadmap: scaling mass production, balancing in-house development with partnerships, and advancing physical AI models tailored to its hardware. Wang said he expects significant breakthroughs by 2027 or 2028, when Unitree robots equipped with physical AI models come online.
Wang also offered a bold prediction: with the arrival of GPT-6, physical AI is showing signs of a breakthrough. He believes embodied intelligence could reach a ChatGPT-like tipping point within one to two years, or perhaps two to three at the outside.
To further build the industry ecosystem, Unitree has partnered with Sequoia China to launch the "Shan Yu · Chuang Plan." This initiative leverages a dual "engineer-plus-investor" perspective to identify and fund early-stage projects in embodied intelligence, robotics applications, key upstream and downstream technologies, and other frontier hard-tech sectors.
Buoyed by recent positive industry news, capital markets have reacted optimistically. Unitree shares closed up 8.74% on October 28 at 695.90 yuan, giving it a market valuation of 281.5 billion yuan. The stock has rebounded more than 50% since the beginning of October.
However, on October 29, Unitree opened lower by 6.13% at 653.22 yuan.
The swift repricing following the earnings release underscores a simple truth: market sentiment and company fundamentals are two very different things.
Going forward, whether Unitree’s momentum can be sustained depends on one thing: shifting its revenue base away from “research and education” toward sustainable, recurring orders from factories and commercial environments.
