Unitree's First Quarterly Report After IPO Shows Revenue Growth Without Profit Growth, R&D Expense Ratio Rises to 19.10%

Unitree's First Quarterly Report After IPO: Revenue Up, Profit Down, Embodied Intelligence Begins Paying for Models and Compute Z Finance November 01, 2026 19:52:43 Updated November 02, 2026 09:57:49

Unitree's first quarterly report after going public puts two speeds of the embodied intelligence industry side by side: robots are selling more, but the profit left from each yuan of revenue is shrinking.

On the evening of October 28, Unitree disclosed its third-quarter report, also its first financial report since listing. In the third quarter of 2026, Unitree's revenue grew by nearly 70% year on year, but net profit attributable to shareholders fell 17.37% year on year and nearly halved quarter on quarter. R&D investment accelerated, selling expenses increased, and competition extended from robot hardware to models, developers and industrial investment. The IPO replenished Unitree's funds, and it is putting that money into the next round of competition faster.

The direction of R&D investment is relatively clear. The company said the new investment mainly involves embodied large models, computing power, R&D teams and new products. This reveals where Unitree stands: it has already built a robot product business that can sell and be profitable; to further enable robots to understand tasks, adapt to environments and continuously complete work, it still needs a set of capabilities requiring sustained investment. Accumulated hardware expertise can lower the starting point of this path, but it cannot waive the bills for models, data and computing power.

Revenue Continues to Grow, Third-Quarter Profit Falls Both Year on Year and Quarter on Quarter

First, look at the cumulative data. In the first three quarters of 2026, Unitree achieved revenue of 1.817 billion yuan, up 55.71% year on year; net profit attributable to shareholders was 387 million yuan, up 270.54% year on year. But after deducting non-recurring gains and losses, net profit attributable to shareholders was 349 million yuan, down 18.78% year on year.

Chart note: The data use the comparative basis of the current third-quarter report; the non-recurring profit for the first three quarters of 2025 is an approximate figure back-calculated from the disclosed year-on-year percentage. Gross margin is calculated from operating revenue and operating costs.

The opposite trends of net profit attributable to shareholders and non-recurring profit are mainly related to last year's low profit base. In the first three quarters of 2025, Unitree recognized about 349 million yuan in share-based payment expenses classified as non-recurring gains and losses, which depressed net profit for the period. After excluding such effects, the profitability in the first three quarters of this year is clearly weaker than what the growth in net profit attributable to shareholders suggests.

If the cumulative data are broken down into individual quarters, the changes become clearer.

Chart note: Q2 is the cumulative first-half figure minus Q1; single-quarter gross margin is calculated based on the revenue and costs of the corresponding quarter, and quarter-on-quarter changes are calculated using unrounded data.

The second quarter of this year was the high point for revenue and profit in the first three quarters. Third-quarter revenue fell 8.82% quarter on quarter, while net profit attributable to shareholders and non-recurring profit fell 49.69% and 48.43%, respectively. Compared with the same period last year, third-quarter revenue grew 69.93%, net profit attributable to shareholders fell 17.37%, and non-recurring profit fell 17.42%. The profit pressure in the third quarter of this year is reflected in both year-on-year and quarter-on-quarter dimensions.

English text is being prepared.

Compared with selling expenses, the acceleration in R&D investment was more concentrated in the third quarter.

In the first three quarters of this year, Unitree's R&D expenses reached 263 million yuan, up 191.48% year on year; R&D expenses as a share of revenue rose from about 7.73% in the same period last year to 14.47%. Third-quarter R&D expenses alone were 127 million yuan, up 249.58% year on year, equivalent to 93.48% of R&D expenses for the entire first half, with a single-quarter R&D expense ratio of 19.10%.

This means that in the third quarter, for every 100 yuan of revenue Unitree generated, about 19 yuan corresponded to R&D expenses. Unitree's original advantages were more reflected in robot hardware, motion control and engineering capabilities. After entering more complex work scenarios, robots also need to understand instructions, recognize environments, plan tasks and adjust actions when the environment changes. This makes models, data and computing power parts that require continuous investment. The fundraising and investment arrangements already reflect this direction. Of the approximately 4.202 billion yuan in project funds originally planned for investment in the prospectus, the "Intelligent Robot Model R&D Project" plans to invest about 2.022 billion yuan, accounting for about 48%.

At the product level, Unitree's UnifoLM-WLA-1.0 model publicly unveiled this year has an officially disclosed parameter scale of 6 billion, was trained on about 2,500 hours of real robot data, and attempts to cover desktop manipulation and whole-body coordination tasks.

Unitree UnifoLM-WLA-1.0 Model Structure

For financial statements, the characteristic of such investment is that expenses occur first and revenue is realized later. Data collection, model training and team expansion require continuous payment, while returns depend on whether robots can enter more scenarios and obtain repeat orders. Unitree's third-quarter R&D expense ratio already reflects the cost of this stage.

Selling Expenses Double, Competition Between Unitree and AgiBot Extends to the Ecosystem

In the first three quarters of this year, Unitree's selling expenses reached 230 million yuan, up 202.45% year on year, a growth rate notably higher than that of revenue; the selling expense ratio rose from about 6.51% in the same period last year to 12.65%. The listing announcement's explanation for the expense growth in the first half included brand promotion activities such as the CCTV Spring Festival Gala, as well as an increase in sales personnel. Judging from cumulative spending, Unitree is investing more resources in brand exposure, customer acquisition and its sales organization.

This is happening in a market where competition has clearly intensified. On October 29, AgiBot co-founder Peng Zhihui disclosed that the company's overall revenue in the first half of this year was close to 2 billion yuan; Unitree's revenue in the same period was 1.152 billion yuan. The two have different product structures, but this revenue scale shows that the rivals Unitree faces already possess considerable commercialization scale.

In terms of shipments, a "two superpowers and multiple strong players" pattern has begun to take shape in the humanoid robot market. Counterpoint statistics show that in the first half of 2026, AgiBot shipped about 9,700 humanoid robots globally, with a share of more than 43%; Unitree shipped more than 7,000 units, with a share of about 31%. Together the two account for about three-quarters of the market, followed by companies such as Galbot, UBTech and Leju.

Source: Counterpoint

Competition has also begun to extend beyond complete-machine sales. AgiBot and Hillhouse have jointly established an embodied intelligence industry fund; the fund was set up in 2025 and completed filing, with an investment scope covering key upstream components, AI algorithms, robot bodies and system integration, as well as downstream application solutions. In April this year, AgiBot founder, chairman and CEO Deng Taihua disclosed that over the past year and a half AgiBot had invested in more than 20 companies.

Unitree, meanwhile, announced on October 28 this year together with Sequoia the launch of the "Shanyu·Innovation Program," aimed at early-stage projects in embodied intelligence, robot applications and key upstream and downstream technologies, combining Unitree's engineering capabilities with Sequoia's investment and corporate service capabilities.

Both leading robot companies are leveraging top financial investment institutions to compete for more technical teams, component companies and application developers. The participation of Hillhouse and Sequoia extends competition from companies' internal R&D and sales further into external project networks and capital resources.

Ecosystem investment and selling expenses belong to different financial calibers, but the business objectives behind them share something in common: getting more companies to develop products around one's own robots and solve customer problems. For the humanoid robot industry, where applications are still being explored, whoever can attract more partners may gain more deployment opportunities, usage feedback and follow-on orders.

Unitree has successfully listed, but the embodied industry is still waiting for commercialization to materialize

Humanoid robot shipments are growing fast, but the application structure is still at an early stage.

According to Counterpoint statistics, in the first half of 2026, entertainment performances, data production, scientific research and other uses together accounted for more than 60% of global humanoid robot shipments, manufacturing accounted for about 13%, and warehousing and logistics about 5%. Scientific research, training and performances are all real demand. However, the procurement logic behind them differs markedly from factories continuously buying more equipment to lower long-term production costs. The former type of demand can be driven by R&D budgets, project budgets and financing; the latter relies more on reliability, cost of use and investment payback periods.

Source: Counterpoint

The capital market's judgment of this stage is also changing. In September this year, Reuters cited people familiar with the matter as saying that regulators had slowed the listing process for some humanoid robot companies through informal window guidance, focusing on high valuations and whether revenue generated by government-supported projects can be sustained. In interviews with Cailianshe during the same period, some leading brokerages also said they had not received notices in the form of window guidance, and that the focus was on tightening the front-end gatekeeping responsibility of sponsor institutions.

The questions from regulators and investors are becoming concrete: who is buying the robots, where the procurement funds come from, whether payment is collected after acceptance, and whether the first batch of equipment will lead to a second batch of purchases once in use. Government and scientific research procurement does not in itself mean there is a problem with revenue; what needs to be verified is customer independence, the reasonableness of transactions and the sustainability of demand.

If the timeline for exiting through listing lengthens, this change will transmit along the financing chain. Investors will need to bear longer holding periods, companies will need to prepare for longer funding cycles, and companies that rely on the next round of financing to sustain R&D and delivery will come under greater pressure. This is the business impact that a change in IPO pace may produce.

Unitree has already "made it ashore" and turned the listing window into actual funds. The company completed its IPO in August this year, raising net proceeds of about 5.917 billion yuan; as of the end of the third quarter, its monetary funds reached 7.354 billion yuan. By comparison, net cash flow from operating activities in the first three quarters was 194 million yuan. Unitree's current cash reserves come largely from listing financing; although this cannot be regarded as a simultaneous improvement in operating capability, it does provide a larger buffer for model R&D, product iteration and market expansion.

An apt detail is that at today's launch event, AgiBot proposed Worken, or "effective intelligent work hours," attempting to measure value using the number of active robots, autonomous working hours and an output-efficiency ratio. When a company leading in shipments also begins to emphasize effective workload, the next scorecard for competition will require more production ledgers from customers.

In the next stage, cash reserves will determine how much time a company can give technology iteration, while real orders and sustained payment collection will determine whether these investments can generate returns.