Which humanoid robotics companies are actually making money?
In our humanoid robotics market deck, you will find everything you need to understand the market
SUMMARY
Unitree is the only major general-purpose humanoid robotics company we can confidently say is making substantial money today. ExRobots has produced a small profit in a specialized niche, while Booster Robotics has reached positive monthly operating cash flow but has not yet proved sustained full-year profitability.
The striking part is how short the profitable list remains. Humanoid shipments are already counted in the thousands, commercial contracts are getting much larger and real factories are using robots for meaningful work, yet revenue has spread through the industry much faster than company-level profit.
Unitree stands out because several pieces of evidence line up at once: RMB1.70 billion of 2025 revenue, positive statutory earnings, positive adjusted earnings, high gross margins and more than 5,500 humanoids shipped during the year. Most competitors can prove only two or three of those things.
Humanoids are no longer a small side business inside Unitree. They grew from roughly 2% of revenue in 2023 to more than half in 2025, making it much harder to argue that the company's profitability simply comes from its older quadruped business.
Unitree's margins probably will not stay as extraordinary as they looked in 2025. Its 2026 updates already show revenue continuing to rise while profit comes under pressure from heavier R&D, sales spending, international expansion and competition.
UBTECH is the clearest counterexample to the idea that large humanoid revenue automatically creates profit. It generated roughly RMB2.00 billion of total revenue in 2025, including about RMB821 million from full-size embodied humanoids and related services, but still lost close to RMB790 million.
AGIBOT may eventually challenge Unitree financially, but the public evidence still stops short. Roughly RMB1.05 billion of reported 2025 revenue and very large production volumes prove commercial scale; without comparable net-income disclosure, profitability remains unproven rather than assumed.
The easiest early humanoid markets are not necessarily the ones most people imagine. Research, education, data collection, entertainment and exhibitions can support sales before robots are capable of replacing a human worker across an entire factory shift.
That helps explain why Booster and ExRobots can reach healthier economics earlier than some industrial humanoid companies. Their customers can justify buying the machines for experimentation, interaction or specialized applications without demanding perfect autonomy, industrial uptime and immediate labor-cost parity.
The broader takeaway is that humanoid robotics has already crossed the commercialization threshold but barely crossed the profitability threshold. Unitree has shown that thousands of humanoids can be manufactured and sold while leaving meaningful earnings behind; nearly every other major player is still trying to prove that last part.
Which humanoid robotics companies are actually profitable today?
Unitree is currently the clearest major humanoid robotics company making real profits, while ExRobots is a much smaller profitable edge case and Booster Robotics has only reached the weaker milestone of positive monthly operating cash flow.
That is a surprisingly short list. Humanoid robot sales are already measured in hundreds of millions of dollars globally, Chinese manufacturers are shipping thousands of units, and US startups have raised billions. Yet most of the companies people associate with the humanoid boom still fall into one of three groups: they generate revenue while losing money, they have commercial pilots without disclosed financials, or they have large orders that will turn into revenue later.
Unitree stands apart because its financial disclosure goes much further. Its Shanghai listing documents show RMB1.70 billion of 2025 revenue and RMB590.75 million of profit attributable to the parent after excluding non-recurring items. More than 5,500 humanoids shipped during that year.
UBTECH, by comparison, produced RMB2.00 billion of revenue in 2025 and had already built a serious humanoid business, yet it still recorded a very large loss. AGIBOT said revenue reached RMB1.05 billion, although there is still no comparable public bottom-line figure. Figure, Apptronik and Agility have genuine industrial deployments but have yet to show evidence of company-level profitability.
So the profitable end of humanoid robotics remains extremely concentrated. Revenue is spreading much faster than profits.
| Company | What is already proven | Profit status today |
|---|---|---|
| Unitree | Large revenue, 5,500+ humanoids shipped, audited financial disclosure | Clearly profitable |
| ExRobots / DS Technologies | Commercial bionic humanoid sales | Small profit in 2025, weaker since |
| Booster Robotics | 700+ deliveries by late 2025 | Positive monthly operating cash flow |
| AGIBOT | RMB1.05B reported 2025 revenue | Profit still unclear |
| UBTECH | RMB2.00B total revenue, large humanoid business | Loss-making |
| Figure / Apptronik / Agility | Real commercial deployments or orders | Profit unproven |
| 1X | Meaningful revenue | Heavily loss-making |
If you want more recent data on this point, please see our latest humanoid robotics market report.
What does “making money” mean for a humanoid robot company?
A humanoid robotics company should count as profitable only when the whole business earns money after its normal costs, because selling robots and making a profit from them are very different achievements.
This distinction changes the ranking immediately.
A signed order shows that a customer wants the robot. Revenue tells us that some of that demand has turned into actual sales. Gross profit tells us the selling price exceeds direct production costs. Company-level profit goes further because it includes the huge R&D teams, AI training, field engineers, sales staff, factories and corporate costs required to keep a humanoid company running.
Agility Robotics gives us a good current example. Its SEC materials disclose more than $300 million of multi-year Digit v5 orders involving roughly 1,000 robots. Agility itself warns investors that this number is not current-period revenue. The robots still need to be deployed and contractual milestones still need to be met.
Booster Robotics illustrates another grey area. The company says monthly operating cash flow turned positive in late 2025. That is encouraging because the operating business brought in more cash than it spent during that month. One good month still tells us far less than a profitable full year.
For this article, we therefore reserve “profitable” for companies where the evidence reaches the company income statement. It is a much tougher test, and most famous humanoid startups currently fail it.
This market map, featured in our humanoid robotics market deck, highlights top companies and startups in the humanoid robotics market
Is Unitree actually profitable?
Yes. Unitree is genuinely profitable, and its latest public financial disclosures make it the strongest large-scale profitability case we can find anywhere in humanoid robotics.
Unitree's revenue rose from RMB159 million in 2023 to RMB393 million in 2024 and RMB1.70 billion in 2025. The business grew more than tenfold in two years.
The profit trajectory moved even faster. Unitree went from a loss in 2023 to RMB94.5 million of reported net profit in 2024 and RMB278.2 million in 2025. After excluding non-recurring items, 2025 profit attributable to the parent reached RMB590.75 million.
That adjusted figure looks unusually high beside statutory net income because the accounting adjustments work differently from the “adjusted EBITDA” figures US startups commonly promote. The useful point is simpler: Unitree reported positive earnings under multiple profit measures, alongside a core-business gross margin of roughly 60%.
Its volumes also put the financial result on a different footing from a pilot-heavy startup. According to Unitree's listing documents, the company sold 5,632 humanoid robots across 2023–2025, with more than 5,500 of them shipped in 2025 alone. It also sold more than 33,000 quadruped robots over the three-year period.
Revenue, shipments, gross profit and company earnings are all pointing in the same direction. Very few humanoid manufacturers currently give us all four.
Is Unitree really making its money from humanoids?
Yes. Humanoid robots have grown into a major part of Unitree's business, so its profitability can no longer be explained mainly by its older robot-dog franchise.
Humanoids contributed only around 2% of Unitree's revenue in 2023. By 2025, their share had climbed above half of company revenue. Unitree was therefore generating roughly RMB850–880 million from humanoid robots in a single year.
That shift happened extraordinarily quickly. In two years, a product category that was financially tiny became Unitree's largest business while the entire company expanded more than tenfold.
The margins make the change even more interesting. Unitree's listing materials put the gross margin of its humanoid business above 60% in 2025. That is an unusually high figure for sophisticated physical hardware, particularly in a market where many competitors are still spending heavily to reduce manufacturing cost.
We still cannot calculate a standalone net margin for Unitree humanoids because the company reports corporate expenses across the broader business. Quadrupeds and other products continue to contribute revenue and profit.
Even with that limitation, humanoids are far too large to treat Unitree's profits as an unrelated legacy business. The company has already built a sizeable profitable robotics operation while humanoids account for most of its new commercial momentum.
As this chart shows, and as featured in our humanoid robotics market deck, search interest in where to buy robots has been rising steadily
Can Unitree keep its unusually high humanoid robot margins?
Unitree should remain profitable for now, although its latest numbers show that the easy part of the margin story may already be ending.
The first warning appeared in the company's 2026 financial updates. First-quarter revenue reached roughly RMB423 million, up about 68% year over year. Adjusted profit attributable to the parent fell more than 50% to around RMB40 million.
That combination is hard to ignore: Unitree was still selling much more while keeping much less of the incremental revenue as profit.
The company pointed to higher R&D, sales and other operating costs. That makes sense given where Unitree is heading. It is hiring, building more capable embodied-AI systems, expanding internationally and preparing for much larger manufacturing volumes. Those expenses arrive well before every investment produces new revenue.
The broader first-half guidance looks healthier than the first quarter, but it tells the same basic story. Unitree expects revenue of roughly RMB1.05–1.13 billion, equivalent to growth of around 36–45%. Adjusted profit is expected around RMB236–283 million, which would be down roughly 6–22% from the comparable period.
Price competition adds another risk. Chinese humanoid manufacturers are entering the market quickly, and several are explicitly pushing lower-cost models. Unitree itself warns in its listing documents that more competition could pressure selling prices and margins.
Unitree's profitability looks real. A 60% gross margin and 30%-plus adjusted net margin simply look much harder to sustain once the humanoid market starts industrializing in earnest.
If you want more recent data on this point, please see our latest humanoid robotics market report.
Is UBTECH profitable now that it sells so many humanoid robots?
No. UBTECH has become one of the largest commercial humanoid robot companies, but the business is still losing hundreds of millions of yuan.
The interesting part is how much commercial progress UBTECH has made without crossing the profitability line.
Total revenue reached RMB2.00 billion in 2025, up more than 50% from the previous year. Revenue from full-size embodied humanoid robots and related services jumped from only RMB35.6 million to about RMB821 million.
Humanoids therefore went from a small experiment to one of UBTECH's main businesses in a year.
Gross margin also improved substantially, reaching about 38% across the company. Yet UBTECH still recorded a net loss close to RMB790 million. R&D cost more than RMB500 million, selling expenses approached RMB470 million and administrative costs exceeded RMB330 million.
The latest half-year results show further improvement without a profit breakthrough. Revenue roughly doubled to RMB1.27 billion, humanoid revenue reached around RMB590 million and company gross margin climbed above 44%. UBTECH still lost roughly RMB339 million.
UBTECH is one of the best reality checks in the entire sector. Customers are buying the robots. Revenue has become large. Gross economics are improving. The cost of building, selling and supporting the whole business remains too high.
UBTECH could reach break-even if humanoid volume keeps climbing while operating expenses grow more slowly. It simply has not reached that point yet.
This chart, featured in our humanoid robotics market deck, illustrates yearly venture capital funding for humanoid robotics startups
Is AGIBOT profitable after passing RMB1 billion in revenue?
We still cannot confidently call AGIBOT profitable, even though it has become one of the world's biggest humanoid robot sellers.
AGIBOT chairman Deng Taihua said 2025 revenue reached roughly RMB1.05 billion, compared with around RMB60 million a year earlier. That would mean revenue grew roughly seventeenfold in twelve months.
Independent market estimates broadly support the scale. Counterpoint ranked AGIBOT first globally by humanoid robot sales revenue for 2025 and estimated its humanoid revenue above $140 million.
AGIBOT also crossed 10,000 cumulative robots produced in early 2026. Management has talked about the company developing stronger internal cash-generation capacity, another encouraging sign.
None of those disclosures give us the number we need most: annual net income.
AGIBOT is private, so its reporting is much thinner than Unitree's listing prospectus or UBTECH's public-company accounts. We do not have a clean annual gross margin, operating profit or audited net profit figure that allows the same judgment.
The conclusion is deliberately narrow. AGIBOT has proved large-scale revenue and manufacturing. Profitability remains unknown. If the company eventually opens its financial statements and shows positive earnings, it could become Unitree's most important challenger in this ranking.
Do Booster Robotics and ExRobots belong on the profitable humanoid list?
Booster Robotics is getting close to our profitability threshold, while ExRobots has actually crossed it before, although both companies operate in much narrower markets than Unitree, Figure or UBTECH.
Booster said it achieved positive monthly operating cash flow for the first time in late 2025. By then, the company had delivered more than 700 humanoid robots to over 200 customers, with more than half of the units shipped internationally.
Its business kept expanding afterward. Booster has reported more than 1,000 cumulative K1 shipments across dozens of countries, while orders accelerated sharply in early 2026.
Booster's route to commercialization is worth noticing. K1 is aimed heavily at universities, researchers, developers and robot competitions. Those buyers do not need the machine to autonomously replace an employee for eight hours a day. They can justify purchasing the platform because they want to experiment with humanoid hardware.
That makes the path to revenue much shorter.
ExRobots goes even further toward a specialized market. Its human-like robots are designed largely for museums, exhibitions, customer interaction and entertainment. Parent company DS Technologies reported roughly RMB74 million of revenue and RMB1.5 million of net profit in 2025.
The profit was tiny, roughly 2% of revenue, and the company slipped slightly back into loss during the first half of 2026 as revenue weakened.
So ExRobots technically belongs among profitable humanoid businesses, while Booster has reached an encouraging cash milestone. Neither gives us the same evidence as Unitree that general-purpose humanoid hardware can be produced profitably at large scale.
If you want more recent data on this point, please see our latest humanoid robotics market report.
This chart, featured in our humanoid robotics market deck, shows how Agility Robotics is capturing share in humanoid robotics
Are Leju Robotics and Kepler making money from humanoid robots yet?
No. Leju Robotics has built meaningful humanoid revenue and Kepler has started winning real orders, but both businesses remain clearly loss-making.
Leju is much further along commercially. Its revenue reached roughly RMB258 million in 2025, up from around RMB56 million a year earlier. Full-size Kuavo humanoids generated approximately RMB178 million from 577 units, which works out to an average selling price around RMB308,000.
The problem appears lower down the income statement. Leju lost roughly RMB70 million during the year, after losses in each of the previous two years. Its gross margin has also moved down from above 50% in 2023 toward roughly 41%.
Customer mix adds an important wrinkle. A large share of Leju's recent revenue has come from data-collection applications and regional robot data centers. Industrial deployments are growing, although they still represent a smaller business than the headlines around humanoid factory workers might imply.
Kepler remains much earlier. Financial information released around Kelin Electric's planned acquisition of control showed only about RMB4.3 million of 2025 revenue alongside a net loss close to RMB67 million.
Confirmed orders were much larger than recognized revenue, reaching more than RMB47 million at one point. That gives Kepler something meaningful to convert into future sales, but the gap between today's revenue and today's cost base is enormous.
Leju has proved people will pay serious money for its humanoids. Kepler has proved that buyers are placing orders. Neither has proved a profitable business model.
Is Figure AI making money from its humanoid robots at BMW?
Figure has proved that its humanoids can perform useful factory work, but we still have no evidence that Figure AI makes a profit from those deployments.
BMW's figures are much stronger than the usual humanoid demo video.
Figure 02 worked in BMW's Spartanburg factory over roughly ten months. The robot accumulated around 1,250 operating hours, moved more than 90,000 components and supported production of more than 30,000 BMW X3 vehicles. It worked ten-hour weekday shifts on a repetitive sheet-metal positioning task.
Those are meaningful operational numbers. Figure has moved beyond a robot walking around a controlled lab.
BMW and Figure have since continued the relationship with the newer Figure 03 platform and additional tasks. That suggests the first program delivered enough value for the customer to keep experimenting.
Financially, however, we know remarkably little. Figure does not disclose BMW contract value, robot pricing, manufacturing cost, field-support cost, gross margin or annual company profit. Private-market reporting around Figure's revenue has also produced conflicting estimates over time.
That gap matters more now because Figure's valuation has climbed into the tens of billions of dollars. A $39 billion private company eventually needs much more than a working robot; it needs an enormous and highly profitable business.
The BMW deployment moves Figure much closer to commercial credibility. It still tells us almost nothing about whether selling or renting Figure robots currently makes money.
This chart, featured in our humanoid robotics market deck, illustrates yearly funding for humanoid robotics startups
Are Agility Robotics and Apptronik already profitable businesses?
Agility Robotics and Apptronik have both built genuine commercial businesses, although neither has shown that the company as a whole is profitable.
Agility now gives us unusually detailed evidence of demand. The company has more than $300 million of multi-year contracted orders for Digit v5, tied mainly to roughly 1,000 robots under three-year arrangements.
Those contracts are serious enough that Agility is preparing to become a public company through a merger valuing the business at $2.5 billion before new capital.
Its operating record is also becoming harder to dismiss. Agility says Digit robots have accumulated more than 65,000 operating hours across commercial deployments with companies including GXO, Schaeffler, Toyota Motor Manufacturing Canada and Mercado Libre.
Still, the investor presentation explicitly says the $300 million order figure should not be read as current revenue. Agility also labels its projected robot margins and payback periods as illustrative economics. Those forecasts may eventually prove accurate, but they are still forecasts.
Apptronik is less transparent financially. CEO Jeff Cardenas has said the company generated more cumulative revenue before its major 2025 fundraising than the total capital it had raised up to that point. Its customer programs include Mercedes-Benz and GXO, and the company has worked with Jabil on manufacturing.
Cumulative revenue exceeding early fundraising sounds impressive but tells us nothing about total expenses. A robotics company can earn $50 million over several years while spending $100 million building the technology.
Agility gives us stronger evidence of contracted commercial scale. Apptronik has a longer revenue history than many people realize. Neither belongs in the profitable column until we see actual earnings.
Are 1X and Boston Dynamics making money from humanoid robotics?
No. The latest available financial evidence points to heavy losses at both 1X and Boston Dynamics despite very different levels of maturity.
1X has started generating much more revenue than it did a year earlier. Norwegian financial accounts reported by Shifter showed sales revenue rising from only around NOK6 million in 2024 to roughly NOK320 million in 2025.
Costs rose much faster. Total costs approached NOK1 billion, leaving an operating loss of roughly NOK649 million.
A thirty- or fiftyfold increase in revenue can look spectacular while the company moves further away from break-even in absolute kroner because engineering, manufacturing and expansion costs are growing even faster.
Boston Dynamics sits at the opposite end of the maturity spectrum. It has been building advanced robots for decades, sells Spot quadrupeds and Stretch warehouse robots, and is developing the electric Atlas humanoid.
Yet financial disclosures connected to parent Hyundai have shown repeated large losses at Boston Dynamics. Reported losses increased from roughly KRW255 billion in 2022 to KRW335 billion in 2023 and around KRW441 billion in 2024.
Those figures cover Boston Dynamics as a whole, so we cannot isolate Atlas. Atlas itself is also much earlier commercially than Spot or Stretch.
Brilliant robotics technology, paying customers and years of product experience can coexist with large company losses for a very long time.
This chart, featured in our humanoid robotics market deck, compares the main business model options for humanoid robot manufacturers
Does Tesla count as a profitable humanoid robotics company because Tesla makes money?
No. Tesla is a profitable company, but Optimus still does not generate a separately reported commercial business, so including Tesla would badly distort the answer.
Tesla already earns billions of dollars from cars, energy products and services. Those profits tell us nothing about whether building Optimus robots makes economic sense.
The company's regulatory filings still discuss Optimus mainly as a product under development and a future manufacturing program. Tesla does not report Optimus revenue as a segment, and there is no disclosed external robot-sales figure from which we could calculate margins.
That will become important once Tesla starts selling meaningful numbers of robots outside its own factories. Internal deployment can prove capability and reduce Tesla's own labor costs, but accounting economics become much clearer when customers begin paying market prices for the product.
For now, Tesla belongs in the “potential future entrant” category when we compare humanoid profitability.
Counting its automotive profit as evidence that Optimus already makes money would be like claiming Amazon's early satellite business was profitable because AWS earned billions.
If you want more recent data on this point, please see our latest humanoid robotics market report.
Are humanoid robot companies actually making money by replacing workers?
Only to a limited extent today. A surprisingly large share of current humanoid revenue comes from research, data collection, education, entertainment and tightly defined industrial tasks rather than fleets of robots replacing full human shifts.
The gap between the popular narrative and the actual customer mix is one of the most useful things the financial data reveals.
Counterpoint estimated that entertainment and performance made up around a quarter of global humanoid sales revenue in 2025. Data-production applications contributed roughly another fifth. Together they represented close to half of the market.
Leju shows the same pattern at company level. A large part of its recent sales came from data-collection projects. Booster sells heavily into research and education. ExRobots makes money from exhibitions and interactive service environments.
Industrial work is becoming more credible at the same time.
Figure's BMW deployment accumulated 1,250 real operating hours. Agility says Digit has surpassed 65,000 hours across commercial sites. UBTECH has been shipping Walker robots into automotive and industrial customers. These deployments have moved well beyond one-off stage demonstrations.
The tasks themselves remain narrower than “replace a worker.” Robots move totes, position components, transport material or repeat a tightly defined handling workflow. Human-built factories are being adapted around them, and field engineers still matter.
That explains how the humanoid industry can already generate substantial revenue while the grand economic promise remains largely ahead of it. Companies have found things customers will pay robots to do today. Full human-level labor substitution is not required to sell the first few thousand machines.
This chart, featured in our humanoid robotics market deck, shows the revenue mix across customer segments in the humanoid robotics market
Why is Unitree profitable while most humanoid robot companies lose money?
Unitree makes money because it combines unusually low hardware costs, high shipment volume and buyers who already have a reason to purchase humanoids without waiting for human-level autonomy.
The manufacturing advantage comes first. Unitree develops important motors, joint modules, structural systems and motion-control technology internally. Vertical integration gives the company more control over its bill of materials and reduces dependence on expensive third-party components.
Volume compounds that advantage. More than 5,500 humanoid shipments in a year spread engineering, tooling, procurement and production overhead across far more units than a typical Western pilot program.
Unitree also sells into a wider set of use cases. Research labs, developers, universities, entertainment customers and industrial users can all buy the same broad hardware platform. The robot does not need to outperform a human factory worker economically before Unitree can recognize revenue.
Compare that with Figure or Agility. Their biggest opportunity lies in replacing or supplementing labor inside large enterprises. Those customers require safety, integration, uptime, support and measurable return on investment. The potential contract value is enormous, but commercial qualification takes much longer.
UBTECH shows that scale alone does not solve everything. Its revenue already exceeds Unitree's, yet R&D, sales and administrative spending still swallow the gross profit generated by the business.
Unitree has found the best economic combination so far: sell a lot of relatively affordable hardware, keep much of the core technology in-house, and serve customers who can use the robot before general-purpose autonomy is finished.
If you want more recent data on this point, please see our latest humanoid robotics market report.
| Company | Commercial strength | Main reason profits differ |
|---|---|---|
| Unitree | Thousands of annual humanoid shipments | High volume, vertical integration, broad buyer base |
| UBTECH | Large and rapidly growing humanoid revenue | Heavy R&D and operating expenses |
| Figure | Strong factory validation | Early deployment scale, financials undisclosed |
| Agility | $300M+ contracted orders | Revenue realization still ahead |
| Booster | 1,000+ reported shipments | Lower-cost research and education market |
Which humanoid robotics companies are actually making money?
Unitree is currently the only major general-purpose humanoid robotics company we can confidently call substantially profitable; ExRobots has produced a small profit in its specialized niche, while Booster Robotics has reached positive monthly operating cash flow without yet proving full-year profitability.
That is the direct answer.
As seen above, Unitree reported roughly RMB1.70 billion of 2025 revenue and RMB590.75 million of adjusted parent profit while shipping more than 5,500 humanoid robots. Humanoids had also grown past half of its revenue. Its latest results show lower profit growth as spending rises, so we should expect margins to normalize, but the company remains in a financial position none of its major pure-play competitors has publicly matched.
ExRobots deserves a footnote-sized place in the profitable group because parent DS Technologies made a small net profit in 2025 from a business centered on realistic bionic humanoids. Its subsequent performance has been weaker, and the company operates in entertainment and service applications rather than the general-purpose labor market.
Booster is the next interesting case. Positive monthly operating cash flow suggests its research-and-education strategy can support healthier economics much earlier than factory-focused competitors. We need a longer financial track record before upgrading that to sustained profitability.
AGIBOT could already be profitable, but its public disclosures still stop short of proving it. RMB1.05 billion of reported annual revenue and very fast shipment growth put it near the top of our watch list.
UBTECH has become a genuinely large humanoid seller and still loses money. Leju, Kepler, 1X and Boston Dynamics are also loss-making based on the latest accounts we can verify. Figure, Apptronik and Agility have increasingly convincing commercial deployments, yet none has publicly demonstrated company-level profit. Tesla makes money as a corporation, while Optimus remains a future commercial business rather than a reported profit center.
The bigger story is that humanoid robotics has already crossed one threshold and barely touched the next one. Customers are buying robots in real volume. Factories are running them for thousands of hours. Contracts are reaching hundreds of millions of dollars.
Profitable manufacturing remains rare.
That makes Unitree more interesting than simply being today's financial leader. It has already solved one of the questions most of the industry is still trying to answer: can a company build humanoid robots, sell thousands of them and have meaningful profit left at the end?
For Unitree, the answer today is yes. For almost everyone else, we are still waiting for the financial statements to catch up with the robots.
If you want more recent data on this point, please see our latest humanoid robotics market report.
This chart, featured in our humanoid robotics market deck, shows how factory humanoid robot technology has evolved over time
OUR METHODOLOGY
This analysis tests which humanoid robotics companies are actually making money rather than simply generating revenue, taking orders, shipping robots or running successful pilots. We separate those milestones because the public discussion around humanoid robotics often mixes them together even though they say very different things about the underlying business.
We reserve “profitable” for cases where the evidence reaches company-level earnings after normal operating costs. Revenue, gross profit, contracted orders, shipments and operating cash flow are still useful, but none of them is treated as a substitute for net income or an equivalent company-level profit measure.
For each company, we looked at several dimensions separately: whether customers were actually buying the robots, whether orders had converted into recognized revenue, what the available margins said about the product economics, whether those economics survived R&D and other corporate costs, how financially important humanoids were inside the broader company, and whether the most recent results strengthened or weakened the profitability case.
We apply an evidence hierarchy. Audited financial statements, regulatory filings, listing documents and stock-exchange disclosures carry the most weight. Customer-side operating records and direct company disclosures come next. Independent market research is used mainly to compare shipment volumes, revenue rankings and application mix across companies.
When a private company reports revenue, orders or production volume without disclosing earnings, we stop where the evidence stops. “Profitability unproven” does not mean we assume the company is losing money; it means the available disclosure does not support a stronger conclusion.
We also separate company economics from humanoid economics. A profitable parent company does not automatically make its humanoid program profitable, which is why Tesla's corporate earnings do not count as proof that Optimus makes money. Conversely, when humanoids become a large share of a robotics company's revenue, as they have at Unitree, that product exposure becomes important evidence even when standalone humanoid net income is not reported.
Specialized humanoid businesses are included, but we distinguish them from general-purpose industrial platforms. Research, education, entertainment, exhibitions and data collection can support a much shorter route to revenue than factory applications that require high uptime, safety validation, integration and measurable labor economics.
The final judgment is therefore not a mechanical ranking based on one number. Company-level profitability carries the most weight, then we test it against commercial scale, humanoid exposure, recent financial direction and disclosure quality. This prevents a large order, an impressive deployment or a fast-growing revenue number from being mistaken for a profitable business.
Key sources include Shanghai Stock Exchange material covering Unitree's listing disclosures and financial performance, UBTECH's official 2025 results filing, UBTECH's investor-relations disclosures, Counterpoint Research's 2025 humanoid commercialization and revenue analysis, Counterpoint's H1 2026 shipment update, and Booster Robotics' company timeline covering deliveries and operating cash flow.
We also used BMW's first-hand operating data on Figure deployments, BMW's subsequent Figure 03 deployment announcement, Agility Robotics' SEC-filed transaction announcement, Agility's SEC-filed investor presentation clarifying its contracted-order figures, Apptronik's Mercedes-Benz commercial agreement, and Apptronik's manufacturing agreement with Jabil.
Additional primary sources used to test specific company claims include Leju Robotics' Shenzhen Stock Exchange prospectus, Tesla's SEC-filed 2025 annual report for the status of Optimus, and Hyundai's audited financial disclosure used for the Boston Dynamics discussion.
In our humanoid robotics market deck, we identify pain points entrepreneurs should prioritize
Related blog posts
- Which humanoid robotics startups are actually making the most money today?
- Which humanoid robotics company has the biggest order book?
