What are the top humanoid robotics startups by revenue?

The ranking of top startups in the humanoid robotics market by revenue

Below is a table ranking all the companies in this market by their current revenue scale. You can find our methodology at the end of this page.

If you want a deeper understanding of the market and its current dynamics, get our humanoid robotics market report.

Which humanoid robotics startups make the most revenue today?

AgiBot, Unitree Robotics and UBTECH currently sit in a clear top tier by the strongest public revenue evidence we could find.

The first surprise is how different this ranking looks from a list based on funding, valuation or media attention. AgiBot reported RMB1.05 billion of company revenue for 2025. Unitree disclosed RMB868 million specifically from humanoid robots, while UBTECH reported RMB820.6 million from full-size embodied-intelligence humanoid products and services. Leju Robotics comes next with RMB178 million from its Kuavo humanoid line.

After Leju, the numbers fall quickly. EX Robots reported RMB54.7 million of relevant product revenue and DOBOT RMB20 million from its embodied-robot category. Richtech Robotics generated $5.05 million across its broader service-robotics business, Agility Robotics recorded about $1.8 million of net sales, and Kepler Robotics disclosed RMB4.34 million.

The ranking becomes much less clean below those companies because many well-known humanoid startups still disclose orders, robots delivered, manufacturing capacity or customer deployments instead of revenue. We keep those metrics separate rather than pretending they are interchangeable.

Rank	Company	Best recent financial evidence	What the figure covers
1	AgiBot (Zhiyuan Robotics)	RMB1.05B	2025 company revenue
2	Unitree Robotics	RMB868M	2025 humanoid revenue
3	UBTECH Robotics	RMB820.6M	2025 full-size humanoid revenue
4	Leju Robotics	RMB178M	2025 Kuavo humanoid revenue
5	EX Robots	RMB54.7M	2025 relevant product revenue
6	DOBOT	RMB20.0M	2025 embodied-robot revenue
7	Richtech Robotics	$5.05M	FY2025 company revenue
8	Agility Robotics	~$1.8M	2025 net sales
9	Kepler Robotics	RMB4.34M	2025 company revenue

Is AgiBot really the biggest humanoid robotics startup by revenue?

AgiBot currently has the largest recent revenue figure we found, although its RMB1.05 billion figure covers the whole company rather than humanoids alone.

Chairman and CEO Deng Taihua disclosed that AgiBot generated RMB1.05 billion in 2025, compared with roughly RMB60 million the year before. That works out to around 17.5 times more revenue in a single year.

AgiBot was founded in 2023, so crossing RMB1 billion this quickly is unusual even in a market growing as fast as humanoid robotics. Commercial activity has also moved well beyond a handful of prototypes. The company has said it ranked first globally in humanoid shipments in 2025, and its production milestones have continued climbing since then.

There is still one important distinction in the ranking. AgiBot sells several types of embodied robots, so the RMB1.05 billion figure cannot be treated as pure humanoid revenue in the same way as Unitree's humanoid segment or UBTECH's full-size humanoid segment.

Even with that limitation, the gap is large enough for us to keep AgiBot first. The revenue figure is current, directly disclosed and supported by substantial manufacturing activity.

How much revenue does Unitree make from humanoid robots?

Unitree made RMB868 million from humanoid robots in 2025, turning humanoids into the company's largest business almost overnight.

Unitree generated RMB1.699 billion of total revenue that year, which means humanoids contributed 51.8%. Two years earlier, humanoids represented just 1.9% of company revenue and quadruped robots were overwhelmingly dominant.

That shift is more revealing than the headline growth rate. Unitree did not simply add a successful new robot to its catalogue. The center of the company moved from quadrupeds to humanoids in two years.

Its IPO disclosures also show more than 5,500 humanoids shipped in 2025. That scale is especially interesting because Unitree operates at much lower price points than several industrial competitors. The company is showing that humanoids can already generate serious revenue through volume rather than relying entirely on expensive enterprise deployments.

Unitree is also one of the rare humanoid companies where the financial evidence extends beyond revenue. Its filings show a profitable company with strong gross margins, something we do not see across most of the market yet.

How much money is UBTECH making from humanoid robots?

UBTECH generated RMB820.6 million from full-size humanoid robots in 2025, putting it almost level with Unitree despite selling far fewer machines.

According to UBTECH's annual report, humanoid revenue jumped from RMB35.6 million the previous year to RMB820.6 million. Full-size humanoids consequently became the company's largest revenue category, accounting for roughly 41% of its RMB2.00 billion total revenue.

UBTECH also reported sales of 1,079 full-size humanoids. That makes this one of the best disclosures we have anywhere in the industry because the company gives us both revenue and unit volume for a clearly defined humanoid category. Few makers give both revenue and unit volume the way UBTECH does; we compare who is ahead in our humanoid robots market report.

The contrast with Unitree is striking. Unitree shipped more than five times as many humanoids, yet the two companies produced almost the same humanoid revenue.

UBTECH is currently selling fewer robots at a much higher average value, which fits its focus on industrial manufacturing and the Walker product family.

Does Unitree or UBTECH make more money from each humanoid robot?

UBTECH appears to generate far more revenue per humanoid than Unitree, even though Unitree sells many more robots.

Using the companies' reported revenue and unit figures, UBTECH's 2025 humanoid revenue works out to roughly RMB760,000 per reported unit. Unitree's disclosure implies less than roughly RMB158,000 per humanoid shipped.

We should not treat those numbers as literal average selling prices. Revenue can include software, services, different product configurations and timing differences between shipment and recognition.

The gap is still far too large to dismiss. It shows that two very different humanoid businesses are already emerging.

Unitree has pushed further into volume with relatively accessible platforms used across research, education and a growing range of commercial applications. UBTECH has concentrated much more heavily on higher-value full-size industrial systems.

Neither approach has clearly won yet. Counting robots alone gives a very different picture from counting revenue.

Metric	Unitree	UBTECH
2025 humanoid units	>5,500	1,079
Approx. revenue per reported unit		~RMB760K
Humanoid share of company revenue	51.8%	~41%
Main commercial profile	Higher-volume	Higher-value industrial
Revenue evidence	IPO disclosure	Annual report
Company profitability	Profitable	Loss-making

Is Leju Robotics already making real money from humanoids?

Leju Robotics is already a meaningful humanoid business, although it remains much smaller than AgiBot, Unitree and UBTECH.

Leju's IPO materials show that its Kuavo full-size humanoid generated RMB178 million in 2025. Kuavo had produced only about RMB13.2 million a year earlier, so revenue increased by more than 12 times.

The company sold 577 Kuavo robots at an average price of roughly RMB308,000. Kuavo also became the core of the business, contributing close to 70% of Leju's total revenue.

What makes Leju especially useful for understanding the market is the unusually detailed breakdown of where those robots went.

Data-collection applications accounted for about 45% of Kuavo revenue. Industrial manufacturing contributed less than 4%. The rest came largely from areas such as research, education and commercial services.

Leju therefore has real humanoid revenue and hundreds of sales, but its numbers also warn us against assuming that every commercially sold humanoid is already working on a factory line.

How concentrated is humanoid robotics revenue right now?

Humanoid robotics revenue is currently concentrated in a remarkably small group of companies.

AgiBot, Unitree and UBTECH together account for roughly RMB2.74 billion of the recent direct revenue evidence in our Chinese company sample. Once Leju, EX Robots, DOBOT and Kepler are added, the identifiable total reaches roughly RMB3.0 billion. Three companies holding almost all the revenue leaves a short list, and we weigh the likely future winners in our report on humanoid robots.

So the first three companies represent about 91% of that pool.

This concentration helps explain why the industry can simultaneously look huge and tiny. There are dozens of funded humanoid companies, prototypes appear constantly and large industrial groups are running trials. Yet only a handful have so far converted that activity into revenue at serious scale.

The gap after the leaders is steep. Leju is already far behind the top three, and another large drop appears below Leju.

For now, humanoid robotics has many competitors but very few scaled sellers.

Why are Chinese humanoid robotics companies making so much more revenue?

Chinese humanoid companies are ahead on revenue mainly because they reached meaningful production and shipment volumes earlier.

Omdia estimated that roughly 13,000 humanoids were shipped globally in 2025 and that Chinese companies dominated those volumes. AgiBot ranked first in Omdia's original count, followed by Unitree and UBTECH.

The exact shipment totals vary depending on the source and definition. Unitree later disclosed more than 5,500 humanoids shipped in its IPO materials, above Omdia's earlier estimate. That kind of revision is normal in such a young market and is one reason we prefer company filings whenever they become available.

That earlier commercial activity now shows up in the revenue numbers.

Are humanoid robot sales really growing this fast?

Yes. Humanoid robot shipments have jumped from a niche market into the low tens of thousands, although estimates are still rough.

Omdia put 2025 global shipments at around 13,300 robots, more than five times the previous year's level. Its ranking was dominated by Chinese manufacturers, with AgiBot, Unitree and UBTECH at the front.

Different research groups produce somewhat different totals because the category itself remains messy. A small research biped, a wheeled upper-body robot and a full-size factory humanoid may all appear in different versions of the same market count.

That uncertainty does not erase the underlying jump in activity. Factories are producing thousands of machines where the industry previously talked mostly about dozens or hundreds.

Company filings are now starting to catch up with those market estimates. Unitree alone reported more than 5,500 humanoids shipped during 2025.

Which humanoid robotics companies are growing revenue the fastest?

UBTECH, AgiBot and Leju all grew at rates that would be extreme in almost any hardware industry.

UBTECH's humanoid revenue increased from RMB35.6 million to RMB820.6 million in one year, a rise of about 2,204%. AgiBot moved from roughly RMB60 million of total company revenue to RMB1.05 billion, around 17.5 times higher. Leju's Kuavo line grew from roughly RMB13.2 million to RMB178 million, more than 12 times higher. Growth of 2,204% and a 17.5-fold jump are the headline numbers, and we study who is growing the fastest in our humanoid robots report.

The small starting bases obviously make those percentages easier to achieve. A company moving from pilots into its first real production year can post extraordinary growth without yet being enormous.

The more interesting part is that several companies crossed into nine-figure or billion-yuan revenue at roughly the same time. We are no longer looking at one outlier startup benefiting from a single large customer.

Richtech offers a useful contrast. Its FY2025 revenue increased 19% to $5.05 million. That would be respectable growth for many industrial businesses, but it looks almost flat beside the expansion happening among the leading Chinese humanoid companies.

Company	Earlier figure	2025 figure	Approx. change
UBTECH humanoids	RMB35.6M	RMB820.6M	+2,203.7%
AgiBot company revenue	~RMB60M	RMB1.05B	~17.5x
Leju Kuavo	~RMB13.2M	RMB178M	>12x
Leju total company revenue	RMB55.5M	RMB258M	~4.6x
Richtech Robotics	$4.24M	$5.05M	+19%
DOBOT embodied robots	RMB3.86M	RMB20.0M	+418.8%
Unitree humanoid mix	1.9% of revenue in 2023	51.8% in 2025	Became largest segment

Who is actually buying humanoid robots today?

Humanoid robot revenue currently comes from a much wider mix of customers than factory videos suggest.

Research institutions, universities, data-collection programs, commercial venues and technology developers still account for a meaningful share of demand.

Leju gives us the clearest breakdown. Nearly 45% of Kuavo revenue came from data-collection applications, while industrial manufacturing contributed less than 4%. Unitree also sells heavily into research and education, especially through comparatively affordable products such as the G1.

UBTECH sits closer to the industrial end of the market. Its Walker robots target automotive, electronics and logistics environments where customers can justify higher spending around specific production tasks.

These days, “commercial humanoid” covers several quite different businesses. A robot sold to a university lab generates revenue just as surely as one deployed in an automotive plant, but the two purchases tell us very different things about how close humanoids are to becoming mainstream industrial labor.

Is Figure AI one of the biggest humanoid robotics companies by revenue?

Figure AI is one of the industry's biggest companies by valuation and ambition, but there still is not enough public revenue evidence to place Figure near the top of a revenue ranking.

Figure's operational progress is much easier to measure.

The company said earlier this year that its BotQ factory had produced more than 350 Figure 03 robots while increasing production from one robot per day to one per hour in less than four months. Figure 03 has also returned to BMW's Spartanburg plant after the previous generation contributed to production involving 30,000 vehicles.

Those achievements show that Figure has moved well beyond a prototype sitting in a lab.

They still do not tell us how much revenue the company recognizes. Some robots can remain inside Figure's own fleet, some may be deployed under commercial agreements with undisclosed economics, and manufacturing output does not automatically equal customer sales.

Until Figure publishes sales, leasing revenue or another comparable financial metric, its exact place in a revenue ranking remains unknown.

Which humanoid robotics startups have the biggest orders right now?

Several humanoid companies now report order books worth tens or hundreds of millions of dollars, making future revenue much larger than current recognized sales suggest.

Agility has disclosed more than $300 million of multiyear orders. ROBOTERA has reported more than RMB500 million of cumulative orders. MagicLab announced a RMB150 million procurement agreement, while Noetix has disclosed more than RMB100 million of contract value.

NEURA Robotics has reported an order book above $1 billion across its broader cognitive-robotics portfolio. We do not treat that as humanoid revenue because the figure also covers other types of robots.

1X belongs in the same forward-looking discussion for a different reason. Its NEO launch booked the company's first-year production capacity of 10,000 robots within five days.

These numbers show that demand is forming well beyond the companies currently publishing large revenue figures. They are still orders, contracts or reservations, so we keep them below recognized revenue when ranking current commercial scale.

Company	Recent commercial evidence	Main limitation
NEURA Robotics	>$1B broader robot orders	Not humanoid-only
Agility Robotics	>$300M multiyear orders	Future revenue
ROBOTERA	>RMB500M orders	Orders, not recognized sales
MagicLab	RMB150M agreement	Procurement commitment
Noetix Robotics	>RMB100M contracts	Order value
1X	10,000 NEOs booked	Preorders / production reservations
Figure AI	Hundreds of Figure 03s produced	Manufacturing output, not revenue

Do 1X's 10,000 NEO preorders mean it already has a huge humanoid business?

1X has one of the strongest consumer-demand stories in humanoid robotics, but 10,000 preorders cannot be treated as recognized revenue.

According to 1X, the entire first year of NEO production capacity was booked within five days of launch. The company put that capacity at 10,000 robots.

NEO's early-access purchase option costs $20,000, while a later subscription option is priced at $499 per month. Buying outright or subscribing is a real business-model choice; we compare which business models are winning in our humanoid robotics market report.

Multiplying 10,000 robots by $20,000 would produce a tempting $200 million headline, but it would be a bad revenue estimate. Customers can choose different payment models, delivery timing matters, cancellations are possible and accounting revenue is generally recognized when contractual obligations are actually fulfilled.

The preorder number is still valuable. It suggests consumer interest can be much larger than the small pilot volumes associated with most home-robot projects.

We just should not turn that demand into revenue before the company does.

Why aren't Tesla and Boston Dynamics near the top of the humanoid revenue ranking?

Tesla and Boston Dynamics are major humanoid competitors, but neither currently provides a clean external humanoid-revenue figure that we can rank alongside Unitree or UBTECH.

Boston Dynamics is now manufacturing the production version of Atlas. The company said its initial Atlas deployments were committed to Hyundai and Google DeepMind, and it has just opened a dedicated Robotics Metaplant Application Center inside Hyundai's Georgia manufacturing operation.

Atlas robots are currently being trained there on real automotive logistics and sequencing work. Hyundai's longer-term plan is to move from validation toward much larger deployments across its manufacturing network.

That makes Boston Dynamics increasingly important commercially, but Atlas revenue remains undisclosed.

Tesla presents an even clearer perimeter problem. Tesla reports tens of billions of dollars from cars, energy and services, but those revenues tell us nothing useful about the current commercial scale of Optimus. Internal factory use, production targets and manufacturing capacity are also different from third-party humanoid sales.

Both companies belong in any serious discussion of the competitive landscape. Neither belongs near the top of a revenue ranking until there is humanoid-specific financial evidence.

Are any humanoid robotics companies profitable yet?

Yes, but profitability is still the exception rather than the norm among companies pushing humanoids at scale.

Unitree is the clearest example. Its IPO disclosures show RMB1.699 billion of total revenue in 2025, a core-business gross margin of about 60%, and substantial net profit.

UBTECH gives us the opposite picture. The company reached RMB2.00 billion of total revenue but still lost roughly RMB790 million, although that loss narrowed materially from the previous year.

Leju also remained loss-making despite its sharp revenue increase, reporting a net loss of around RMB70 million for 2025. Richtech's SEC filings likewise show that a few million dollars of revenue is nowhere near enough to absorb the operating costs of building and commercializing advanced robots.

Revenue scale is starting to separate the market, but it does not tell us on its own which business model will eventually produce attractive economics.

Unitree currently stands out because it has combined large unit volumes, meaningful humanoid revenue and profitability.

Which humanoid business model is making more money: cheap robots or expensive factory robots?

Both models are already producing meaningful revenue, and the current leaders suggest there may be room for more than one winner.

Unitree shows what happens when prices come down enough to broaden the customer base. Research labs, universities, developers and companies can buy humanoids without making a giant industrial-capital decision. High volumes then compensate for lower revenue per robot.

UBTECH is following almost the reverse route. Its full-size Walker robots are aimed much more directly at industrial customers, with far higher implied revenue per machine.

Leju sits somewhere between them. Kuavo sells at a mid-range average price and reaches several customer groups rather than depending entirely on factory automation.

What we have not seen yet is enough mature data on recurring software revenue, servicing costs, robot longevity and customer retention to know which model creates the best long-term economics.

Right now, Unitree has the strongest evidence that volume can work financially, while UBTECH shows that industrial customers will pay far more for systems built around demanding production environments.

Is humanoid robotics already a real commercial market, or is it still mostly pilots?

Humanoid robotics is commercially real today, but large-scale deployment is still concentrated in a small part of the market.

The evidence has moved well beyond demos. Companies now report billion-yuan revenue, thousands of robots shipped, large order books and robots working inside real customer operations.

Yet many flagship programs are still small compared with the size of the industries they want to automate. Flagship programs still look small next to the industries they want to automate; we look at the biggest opportunities in our humanoid robots market report.

Figure's BMW work is meaningful, but it remains one customer program. Boston Dynamics is training Atlas inside Hyundai's manufacturing environment before larger deployments. Agility has real production-floor experience but only recently recorded low-single-digit millions of dollars in annual sales.

The market sits in an awkward middle stage. Customers are paying, production lines exist and deployments are happening, while widespread fleet rollouts are still unusual.

Humanoid robotics has moved faster commercially than skeptics sometimes assume, but not nearly as far as the biggest production forecasts imply.

Which humanoid robotics companies could jump into the revenue top tier next?

Agility Robotics, ROBOTERA, Figure AI, 1X and NEURA Robotics currently have some of the clearest paths toward a much larger commercial business, although their evidence is very different.

Agility already has a substantial contracted pipeline that can convert into recognized revenue as Digit deployments expand. ROBOTERA has disclosed a large monetary order book and is scaling deliveries.

Figure has built manufacturing capacity unusually quickly and continues to deepen its BMW deployment. Its missing piece for our purposes is financial disclosure rather than evidence that robots exist and can be produced.

1X has consumer demand that few humanoid companies can match if its preorder book survives the transition into actual deliveries. NEURA has a very large broader robotics pipeline, but we would need cleaner humanoid attribution before putting that figure beside pure humanoid businesses.

Apptronik is another company we would watch closely. It has major industrial partners and several verified deployments, but public commercial numbers remain too sparse to tell us how large the business is today.

These companies could move quickly once orders start showing up as revenue. Until then, we prefer leaving the gap visible.

Is humanoid robotics already a billion-dollar revenue market?

Yes. The revenue we can already document is comfortably large enough to say humanoid robotics has become a real business category.

The direct figures in our dataset add up to several billion yuan even before we include companies that disclose no usable revenue at all.

The bigger change is how the market can now be measured. A few years ago, humanoid companies were mainly compared through technical demos, funding and promised production capacity. Today we can increasingly compare actual sales, product mix, shipment volume, margins and losses.

That gives us a much better way to separate companies that are commercially scaling from companies that are still preparing to do so.

What should we trust more: humanoid revenue, orders or robots shipped?

For measuring commercial scale today, we trust recognized revenue first, then strong financial proxies, then orders and delivered units when better evidence does not exist.

The reason is practical. Revenue tells us that a product or service has progressed far enough through a commercial transaction to appear in the company's accounts.

Orders are useful, especially when customers and contract values are clear, but delivery may stretch across several years. A shipment count tells us manufacturing has happened, yet it says little about price, ownership or contract structure. Production capacity is further removed again.

Freshness still matters. A recent and clearly defined order book can tell us more about where a company is heading than a five-year-old revenue number. We simply keep the two metrics labeled correctly rather than quietly turning one into the other.

That approach is especially important in humanoid robotics because the gap between a company's valuation, announced pipeline, production target and current revenue can be enormous. The gap between valuation and current revenue can be huge; we look at what each company is worth, then and now, in our humanoid robots report.

So who is actually winning the humanoid robot revenue race right now?

AgiBot, Unitree and UBTECH currently lead the humanoid robot revenue race, with Leju forming the clearest second tier behind them.

AgiBot has the largest recent company-revenue disclosure, while Unitree and UBTECH provide cleaner humanoid-specific numbers. Unitree also combines scale with profitability, and UBTECH is generating almost comparable humanoid revenue from a much smaller number of higher-value machines.

Leju has already built a real commercial humanoid business but remains far below the first three. Below Leju, direct revenue evidence becomes much smaller or harder to isolate.

The most striking result is who does not lead this ranking. Figure AI, 1X, Apptronik, Agility Robotics, Boston Dynamics and Tesla are among the industry's most important names, yet their public evidence currently leans much more heavily on deployments, orders, preorders, manufacturing or future production plans.

That could change quickly. Humanoid robotics is moving fast enough that today's order book can become next year's revenue base.

For now, the revenue leaders are the companies that have already turned robots into substantial sales rather than the companies carrying the biggest valuation or the loudest expectations.