Unitree Technology's First Quarterly Report After Going Public: Revenue of 1.817 Billion Yuan, Non-Recurring Profit Down 18.8%

Leiphone

On the evening of October 28, Unitree disclosed its first third-quarter report since going public. In the first three quarters of 2026, revenue already exceeded the full year of 2025, with the company achieving revenue of 1.817 billion yuan, but gross margin fell noticeably, the R&D expense ratio rose sharply, and non-recurring profit actually declined.

Unitree's revenue is still accelerating, but profit has not kept up. On the surface, this is a financial report of "revenue growth without profit growth." In essence, this is Unitree's strategic shift, actively using the hardware dividends earned over the past two years to subsidize embodied large models and computing power investment. Short-term pressure on profit and cash flow is the inevitable cost of this shift. 01

Looking at Unitree through its financial report

Revenue growing, profit under pressure, R&D ramping up

The most noteworthy information in this third-quarter report can be summarized into three core conclusions.

Conclusion one: Revenue is still growing rapidly, but it has shifted from explosive growth to steady volume expansion, and revenue by product line is no longer transparent.

Unitree Technology's operating revenue in the first three quarters of 2026 was 1.817 billion yuan, up 55.7% year on year; third-quarter revenue alone was 665 million yuan, up 69.9% year on year. First-three-quarter revenue has already exceeded the full-year 2025 figure of 1.699 billion yuan.

These figures show that the company's business scale is still expanding rapidly and revenue continues to reach new levels. But the year-on-year growth rate has fallen from 332.7% in 2025 to 55.7% in the first three quarters.

However, this is typical steady growth from a high base, not a decline.

Looking at a longer cycle, Unitree Technology's revenue was only 159 million yuan in 2023, rising to 393 million yuan in 2024 (+146.8%), and soaring to 1.699 billion yuan in 2025 (+332.7%).

2024–2025 were the two years of penetration dividends as humanoid robots rapidly scaled from almost zero. The growth rate decline in 2026 is a normal phenomenon after the base was raised, and it also means the company is moving from "new product dividends" into a "scaled operations" stage.

In this financial report, Unitree Technology attributed revenue growth to product innovation and business expansion. After going public, Unitree continued to launch iterative products. In September, it released the new-generation humanoid robot G1+, optimizing the full-joint motor system and multi-sensor fusion for pain points in real-world deployment, and adding an external power supply interface.

The new product both continues the hardware engineering advantage and provides a new growth point for revenue.

Unitree Technology has also recently continued to launch new products. On October 21, Unitree Technology released the H2-X, a full-size high-dynamic humanoid robot with 31 degrees of freedom, officially rated for up to 8 hours of battery life, targeting emergency rescue and special scenarios.

In the official demonstration video, the H2-X showed high-dynamic, high-load movements such as backflips, airborne kicks, a stable landing from a 4-meter-high platform, denting a car door, and lifting a car for simulated rescue, clearly targeting special scenarios such as security and emergency rescue.

However, the third-quarter report no longer discloses revenue by product line. Its earlier prospectus and annual report data had revealed that in 2025 humanoid robot revenue was 868 million yuan, accounting for 51.1%, overtaking quadruped robots' 698 million yuan for the first time.

Conclusion two: Gross margin has peaked and is falling, and the days of making money through a high-gross-margin product mix may be over for now.

The gross margin in the first three quarters was 54.0%, further falling to 50.6% in the third quarter alone. Compared with 60.4% for the full year of 2025, the decline is obvious.

Gross margin directly reflects product mix, pricing power, and cost control. A gross margin above 50% is still relatively high for a hardware company, but the downward trend shows that the core driver that previously pushed gross margin higher is weakening.

Unitree Technology's gross margin was 44.7% in 2023, rose to 57.2% in 2024, and reached a peak of 60.4% in 2025. The core logic behind this was the structural dividend from the continuously rising share of high-gross-margin humanoid products. In 2025, the humanoid revenue share exceeded half for the first time, directly pushing up the overall gross margin.

The gross margin decline in 2026 is the result of multiple factors: yield fluctuations during the ramp-up of humanoid mass production, price cuts on some models, rising BOM costs for computing power and core hardware, and a possible increase in the share of low-gross-margin industry application orders.

Simply put, the stage in which gross margin could naturally rise by "selling higher-gross-margin products" has ended. The company must truly bring costs down through scale, or rely on software/services to increase added value.

In the first three quarters, net profit attributable to shareholders was 387 million yuan, up 270.5% year on year; but non-recurring net profit attributable to shareholders was 349 million yuan, actually down about 18.8% year on year. In the third quarter, net profit attributable to shareholders was 113 million yuan, down 17.4% year on year; non-recurring net profit was 105 million yuan, down 17.4% year on year.

Net profit attributable to shareholders looks good, but non-recurring data better reflects true operating profit. The huge contrast between the two shows that the growth in net profit attributable to shareholders largely benefited from the low base in the previous year.

In 2025, the company recognized a large share-based payment expense, lowering that year's base for net profit attributable to shareholders, which significantly amplified the year-on-year growth rate in 2026. Non-recurring profit, which truly reflects operating quality, has already begun to come under pressure.

At the same time, Unitree's R&D investment increased substantially. R&D expenses reached 263 million yuan in the first three quarters, up 191.5% year on year. The R&D expense ratio jumped from a stable range of 8.8%–9% in 2023–2025 to 14.47%, and in the third quarter alone it was as high as 19.1%.

The company explicitly stated that the increase mainly came from R&D investment in the embodied intelligence field, especially the substantial growth in model R&D and computing power investment related to embodied large models, as well as the expansion of the R&D team.

This is highly consistent with Unitree's technology roadmap in recent years: in September 2025 it open-sourced the world model UnifoLM-WMA-0, in January 2026 it open-sourced the vision-language-action model UnifoLM-VLA-0, and in September 2026 it officially released and open-sourced the 6-billion-parameter world-language-action model UnifoLM-WLA-1.0.

UnifoLM-WLA-1.0 was trained on approximately 2,500 hours of high-quality real-robot data, enabling a single model to coordinate desktop manipulation and whole-body movement, supporting cross-task and cross-end-effector generalization, and performing leadingly in multiple embodied reasoning benchmarks.

During the same period, Unitree Technology also demonstrated the humanoid robot driven by UnifoLM-X2-1.0 performing continuous offense and defense without remote control.

At the World Physical AI Conference on October 22, Wang Xingxing further revealed that the company is conducting preliminary research on a physical AI robot self-evolution V1.0 large model, and has already deployed GPT-6 Astra on robots in a zero-training manner, completing tasks such as painting, playing Go, and navigating unfamiliar scenes.

Overall, from 2023 to 2025, the company's R&D investment was relatively restrained, mainly focused on the hardware itself, with a stable expense ratio, and it relied on high-margin hardware to turn profitable and achieve substantial profits. In 2026, the company explicitly bet its resources on embodied large models and computing power. This is a proactive choice: using the hardware profits accumulated in the past and the cash-cow position of the quadruped business to exchange for long-term physical AI barriers.

Short-term profits are compressed; for Unitree, this is the inevitable cost of a strategic shift. 02

After going public, it must prove that technology investment can bring returns

Unitree listed on the STAR Market on August 19, 2026, with an issue price of 150.80 yuan. On the first day, it opened and immediately surged to 1,100 yuan, an increase of more than 629%, and its market value once exceeded 440 billion yuan. It then continued to decline, touching a low of around 404 yuan intraday on October 15, a drawdown of more than 60% from the high, with about 280 billion yuan of market value evaporating.

Starting from October 16, the stock price rebounded rapidly. By around the release of the third-quarter report, it had cumulatively risen nearly 70%, returning to around 670 yuan, and market value recovered to about 270 billion yuan. The trailing price-to-earnings ratio remained as high as about 465 times, far exceeding the industry average.

The stock price's sharp rise and fall ultimately reflect the market being pulled back and forth between "long-term imagination" and "short-term performance."

The frenzy in the early days of listing reflected the market's high expectations for the humanoid robot and embodied intelligence track; the subsequent sharp drawdown was the result of extremely high valuations reverting, doubts about early commercialization, and a relatively small free float acting together.

In the near term, the rebound was catalyzed by the release of the new H2-X product, the World Physical AI Conference, upward revisions of institutional ratings, and the industry's scarcity, among other factors.

Wang Xingxing's public statements corroborate the financial report data. He has repeatedly emphasized that the company is increasing investment in embodied large models and physical AI, and believes that the key to future competition lies in the alignment capability between models and real robots, as well as the construction of a data closed loop.

At the conference on October 22, he explicitly stated that physical AI has shown signs of being broken through, and that a breakthrough may come in as soon as one to two years, or at most two to three years. These statements are very consistent with the direction in the third-quarter report of a substantial rise in the R&D expense ratio and a shift in investment focus toward the UnifoLM series and computing power.

Under profit pressure, Unitree still proactively directs resources toward the areas it believes will determine long-term victory or defeat.

Taken together, Unitree's current development direction is already relatively clear: in the short term, it will continue to rely on the scale-up of quadruped and humanoid hardware to maintain revenue growth, while investing substantial resources in embodied large models, world models, and data closed loops, striving to build barriers at the physical AI level. This means that in the coming quarters, profit margins and cash flow may still come under pressure, but revenue scale is expected to continue expanding.

The impact of this third-quarter report on the market is two-way. On the one hand, it uses real data to prove that the company is still growing rapidly and has sufficient cash to support long-term investment, reinforcing its position as a scarce A-share embodied intelligence target; on the other hand, the decline in gross margin, the drop in non-recurring profit, and the deterioration in cash flow quality also remind the market that high valuations correspond to strong expectations for the future, and any commercialization progress falling short of expectations could trigger a new round of valuation fluctuations.

Whether Unitree can endure this painful period depends on two things: whether revenue can gradually shift from being dominated by research and education orders to industrial and commercial deployment; and whether investment in embodied models can quickly turn into visible product capabilities and order premiums. A single-quarter profit figure is only short-term fluctuation; these two things can determine its long-term value more than any short-term profit fluctuation. ▼ Recommended Reading ▼