Which robotics startup is growing the fastest?
In our robotics market deck, you will find everything you need to understand the market
SUMMARY
AgiBot is the fastest-growing major robotics startup we can identify from public commercial evidence today.
The strongest part of the case is that several different measures are moving together. AgiBot is not just raising money or announcing factory capacity: its revenue, production and estimated humanoid shipments have all accelerated sharply.
Its revenue curve is unusually steep for a hardware company. AgiBot went from more than RMB60 million in 2024 to RMB1.05 billion in 2025, while recent reporting says first-quarter 2026 revenue alone exceeded RMB1 billion.
The factory ramp backs up the financial story. AgiBot moved from 5,000 cumulative robots to 15,000 in about six months, with its monthly production pace rising roughly fivefold from the earlier part of the ramp.
Two independent research firms now put AgiBot first worldwide in humanoid shipments. Their exact totals differ, but both arrive at essentially the same conclusion: AgiBot has overtaken Unitree on current humanoid volume.
The important caveat is that a shipped humanoid is not automatically a productive industrial worker. Research, entertainment, data collection and controlled commercial uses still account for a large share of the market, so unit volume is a much looser measure than repeat factory orders.
That is why AgiBot's biggest unanswered question is revenue quality rather than growth speed. We still know relatively little about margins, customer concentration, repeat purchases, or how much demand comes through government-backed projects, investors and ecosystem partners.
Skild AI is the most credible software-first challenger because one intelligence layer can potentially scale across ABB, Universal Robots, Fetch systems and other machines without manufacturing every robot itself. FieldAI has a similar path through Caterpillar, Boston Dynamics and other platforms.
Figure AI has made the strongest recent US humanoid manufacturing jump, while Agility Robotics has cleaner proof that a small number of humanoids can perform the same paid warehouse workflow over and over. Locus Robotics is already a much larger recurring-revenue business, but its growth rate is far more mature.
The next phase of the race will probably look different from the current one. Shipping thousands of robots got AgiBot into first place; staying there will require customers to expand pilots into much larger fleets because the machines reliably save money.
This market map, featured in our robotics market deck, highlights top companies and startups in the robotics market
Why is the fastest-growing robotics startup so hard to identify right now?
The fastest-growing robotics startup is hard to identify today because the companies raising the most money are no longer the same companies shipping the most robots.
Humanoid robotics makes that split especially obvious. Smart Analytics Global estimates that worldwide humanoid shipments jumped from about 5,100 units in the first half of 2025 to 19,100 in the first half of 2026. Counterpoint Research uses a somewhat broader methodology and puts the latest total above 22,000. Either estimate tells us that the market has grown roughly fourfold in a year.
Most of those machines now come from China. Smart Analytics Global estimates Chinese vendors supplied more than 97% of worldwide humanoid shipments during the first half of 2026, with AgiBot moving into first place ahead of Unitree.
Several leading US humanoid startups, by contrast, are being valued partly on what their technology might become. AgiBot is showing something more immediate: a very young company turning robots into revenue and manufacturing volume at an unusually fast pace.
That gap between expected scale and current scale is the real reason this question has become interesting.
What should “growing fastest” actually mean for a robotics startup?
For robotics startups, we think “growing fastest” should mainly mean that the business itself is expanding quickly through revenue, paid deployments and robots reaching customers.
Valuation growth belongs in the discussion, but it cannot lead it. A startup can move from a $2 billion valuation to $20 billion because investors suddenly believe its future market is enormous while its actual revenue remains tiny.
Manufacturing capacity has the same problem. Saying a factory will eventually build 10,000 robots tells us much less than actually delivering thousands.
We therefore care most about revenue growth when credible figures exist. Robot shipments and active deployments come next. Production growth helps confirm that a hardware company can support those sales. Customer expansion becomes particularly useful when pilots turn into repeat orders or larger fleets.
Funding still tells us something. Robotics is expensive, and raising $1 billion can radically increase a company's ability to hire, manufacture and collect data. We simply treat that money as fuel for future growth rather than proof that the growth has already happened.
That also lets us compare very different companies. AgiBot sells hardware. Skild AI sells robot intelligence. FieldAI puts autonomy software onto other companies' machines. Locus Robotics rents warehouse robots as a service. They do not need identical business models to be compared; they need evidence that customers are spending substantially more money with them.
| What we measure | What it tells us | How much weight we give it |
|---|---|---|
| Revenue growth | Customers are spending more money | Very high |
| Shipments and paid deployments | Products are reaching real users | Very high |
| Repeat fleet expansion | Customers found enough value to buy more | Very high |
| Production growth | The company can physically support demand | High |
| Customer partnerships | Future business may be forming | Medium |
| Funding | The company has more capacity to grow | Medium |
| Valuation | Investors expect much bigger future results | Low |
If you want more recent data on this point, please see our latest robotics market report.
As this chart shows, and as featured in our robotics market deck, search interest in robot costs has increased significantly
Why has AgiBot suddenly become the robotics startup to beat?
AgiBot has become the company to beat because several different measures of growth have started pointing in the same direction at once.
Until fairly recently, AgiBot could still be dismissed as one more heavily funded Chinese humanoid company racing toward mass production. That description no longer fits very well.
Two independent research firms now place AgiBot first in worldwide humanoid shipments. Its founder has disclosed more than RMB1 billion of annual revenue. Recent reporting from 36Kr and Caijing, citing investors and people close to the company, says first-quarter 2026 revenue alone exceeded RMB1 billion.
AgiBot has also moved from 5,000 cumulative robots to 15,000 in about six months.
None of those measures is perfect individually. The first-quarter revenue figure has not been published in audited accounts. A produced robot is not necessarily a robot doing productive work. Humanoid shipment definitions vary between research firms.
Taken together, though, the picture is unusually hard to dismiss. We are looking at fast revenue growth, fast production growth and rising market share at the same time.
That combination is currently much rarer than a big robotics funding round.
How fast is AgiBot's revenue actually growing?
AgiBot's revenue growth is extreme even after we ignore the almost meaningless percentage increase from its tiny first year.
Founder and chairman Deng Taihua disclosed that AgiBot generated about RMB300,000 in 2023, more than RMB60 million in 2024 and RMB1.05 billion in 2025.
The useful comparison starts in 2024, once the company was already producing tens of millions of renminbi in sales.
Revenue went from roughly RMB60 million to RMB1.05 billion in one year. That is about 17.5 times as much revenue, or growth of roughly 1,650%.
Then came a more surprising number.
36Kr and Caijing have separately reported, citing investors and people at or close to AgiBot, that the company generated more than RMB1 billion during the first quarter of 2026. We should be more cautious with that figure because AgiBot itself has not released a quarterly financial statement confirming it.
Still, if the reporting is accurate, AgiBot almost matched its entire 2025 revenue in three months.
The implied annual run rate is already above RMB4 billion, which happens to sit around the full-year target reported by 36Kr. Deng has publicly set an even larger goal of more than RMB10 billion for 2027.
Skild AI's move from zero live revenue to roughly $30 million in a few months is arguably faster from a standing start. FieldAI has reportedly accumulated more than $100 million in revenue and customer contracts. Locus Robotics is already a larger recurring-revenue operation.
AgiBot still has the steepest proven revenue curve once we require the company to have reached a meaningful commercial base.
| Period | AgiBot revenue | Change |
|---|---|---|
| 2023 | ~RMB0.3M | Starting year |
| 2024 | >RMB60M | ~200× |
| 2025 | RMB1.05B | ~17.5× |
| Q1 2026, reported by 36Kr and Caijing | >RMB1B | Nearly 2025 revenue in one quarter |
| 2026 reported target | ~RMB4B | Around 4× 2025 if reached |
This chart, featured in our robotics market deck, shows annual venture capital investment in robotics startups
Is AgiBot shipping robots as fast as its revenue is growing?
AgiBot's shipment and production numbers are now moving quickly enough to support the revenue story rather than contradict it.
The company's production milestones show how sharply the factory ramp changed.
AgiBot took roughly a year to go from its first 1,000 robots to 5,000. It then needed only about three months to reach 10,000 and another three months to reach 15,000.
Put that into a monthly production rate and the jump becomes easier to see. The 1,000-to-5,000 stretch worked out to roughly 330 additional robots per month. The next two stretches ran at around 1,650 to 1,700 per month.
So production speed increased roughly fivefold and then held around that higher level for another quarter.
Independent shipment estimates broadly fit those factory numbers.
Smart Analytics Global estimates AgiBot shipped about 8,400 humanoid robots during the first half of 2026, up 562% from the same period a year earlier. Counterpoint Research estimates around 9,700.
The methodologies differ, which is why we should not pretend the exact number is settled. What matters more is that two independent datasets both put AgiBot at roughly nine thousand first-half shipments and rank it first worldwide.
For a three-year-old robotics company, that is already an unusual level of physical scale.
Are AgiBot's robots actually doing useful work?
AgiBot is clearly selling real robots, but we still cannot assume that every shipment represents economically useful automation.
This is the biggest weakness in the headline growth numbers.
Counterpoint Research estimates that entertainment and performance together with data production and research still represented more than 60% of worldwide humanoid shipments during the first half of 2026. Intelligent manufacturing accounted for about 13%, while warehousing and logistics contributed around 5%.
Smart Analytics Global presents a more optimistic picture, saying industrial and commercial applications represented more than 70% of shipments, up from about 50% one year earlier.
The two reports sound contradictory, but their categories differ enough that both can be true. A robot collecting industrial training data, working in a commercial exhibition or assisting in a controlled industrial setting can fall into different buckets depending on the research methodology.
In other words, “humanoid shipment” is still a very loose measure of productive adoption.
AgiBot also sells more than one type of humanoid. Its portfolio includes large bipedal A-series machines, smaller X-series robots and wheeled G-series systems. Some go into manufacturing and logistics. Others serve education, research, customer interaction, entertainment or data collection.
We give AgiBot full credit for manufacturing and selling thousands of machines while giving much less credit to the idea that all of those machines already replace human labor.
The better test will come from repeat purchases. A factory that buys five robots for evaluation and later expands to 200 gives us much stronger evidence than 40 different customers each buying one machine.
That evidence is still developing.
If you want more recent data on this point, please see our latest robotics market report.
This chart, featured in our robotics market deck, breaks down Figure’s playbook in robotics
Did AgiBot really overtake Unitree?
AgiBot has genuinely overtaken Unitree on the latest humanoid shipment rankings, although Unitree remains the more mature robotics business in several important ways.
Smart Analytics Global gives AgiBot 44% of global first-half humanoid shipments and Unitree 31%. Counterpoint arrives at almost exactly the same share split despite estimating a larger total market.
That puts AgiBot's current volume around 40% above Unitree's by either methodology.
Unitree should not be underestimated. Its 2025 revenue reached RMB1.708 billion, comfortably above AgiBot's RMB1.05 billion. Unitree was also profitable before its IPO, something AgiBot has not demonstrated publicly.
The comparison has become cleaner because Unitree is now a listed company. Its shares began trading on Shanghai's STAR Market after an IPO that valued the company at roughly RMB61 billion at the offer price.
Public filings show Unitree generated about RMB423 million of revenue in the first quarter of 2026.
Recent reports put AgiBot above RMB1 billion in the same quarter. That would mean AgiBot was already producing more than twice Unitree's quarterly revenue, although we have to keep the disclosure gap in mind: Unitree's figure comes from public-company reporting, while AgiBot's comes from people cited by financial media.
So Unitree still gives us cleaner financial information and proven profitability. AgiBot currently has the faster trajectory.
The distinction also matters for our title. Unitree is now a public company, so it no longer fits comfortably inside a ranking of private robotics startups.
Is AgiBot's growth too dependent on China?
AgiBot's growth is still heavily tied to China, and that makes its current numbers a little less transferable to other robotics markets.
China currently provides an unusually favorable environment for scaling embodied AI.
The country has a dense network of motor, actuator, battery, sensor, machining and electronics suppliers. It also has the world's largest industrial-robot market, huge automotive and electronics manufacturing sectors and strong local-government enthusiasm for humanoid robotics.
That helps explain why Chinese companies supplied more than 97% of global humanoid shipments in Smart Analytics Global's latest first-half estimate.
AgiBot has another advantage: an unusually broad network of investors, suppliers, local partners and joint ventures. Caijing reported that the company works with more than 400 commercial partners and identified more than ten state-backed projects involving AgiBot directly or indirectly, with several worth more than RMB10 million.
The same Caijing investigation adds an important caveat. People close to the company said some investors and ecosystem partners also purchase robots in connection with broader partnerships.
Those are still sales, but the economics are different from an unrelated factory manager deciding that AgiBot gives the best return on investment and ordering 500 machines.
This is one reason we would like to see more information on customer concentration, repeat orders and revenue mix before treating every renminbi of AgiBot revenue equally.
AgiBot is trying to reduce the domestic dependence. Deng has said the company wants overseas business to contribute around 30% of revenue during 2026 and eventually more than half. Products are already being pushed into Europe and other Asian markets.
For now, though, China's industrial ecosystem remains a major part of why AgiBot can move this quickly.
If you want more recent data on this point, please see our latest robotics market report.
This chart, featured in our robotics market deck, shows annual funding in robotics startups
Could Skild AI be growing faster without building its own robots?
Skild AI is the strongest challenger to AgiBot if we care more about the growth of robot intelligence than the number of robots manufactured.
Skild's strategy is fundamentally different. It is trying to build one general-purpose “brain” that can control many kinds of machines rather than designing one robot around one body.
Commercially, the early ramp is impressive.
Skild says live revenue went from zero to about $30 million within a few months in 2025 and continued growing with multiple customers. The company subsequently raised $1.4 billion at a valuation above $14 billion.
The more interesting developments have come after that financing.
Skild announced deployments with ABB Robotics, Universal Robots and Mobile Industrial Robots. Nvidia said the technology is also being used with Foxconn on high-precision assembly for Blackwell production.
Skild then bought Zebra Technologies' robotics automation business, the operation previously known as Fetch Robotics. That acquisition gives it existing warehouse software, customers and installed automation rather than forcing the company to build every commercial relationship from scratch.
Lately, Skild has also introduced S1, a robotics foundation model designed to learn new tasks from an example without conventional retraining. Technically, that pushes the company closer to the idea behind its enormous valuation.
The commercial problem is simpler: we still do not know the current revenue number.
A move from zero to $30 million is exceptional. Without a later figure, we cannot tell whether that initial ramp became $40 million, $100 million or several hundred million.
Physical Intelligence sits even further toward the research end of the same race. It has raised enormous amounts of capital to build general robot models, but public evidence still focuses far more on model capability than revenue or production deployments.
Robot-brain companies could eventually grow much faster than hardware manufacturers. Skild is the one we would watch most closely. As of now, the public numbers do not let it take first place.
Which US humanoid startup is closest to AgiBot: Figure AI or Apptronik?
Figure AI is much closer to real production than it was a few months ago, while Apptronik is building a serious deployment network, but neither company has shown AgiBot-level volume yet.
Figure has made the bigger manufacturing jump.
The company says it produced more than 350 Figure 03 robots and increased its manufacturing rate from one robot per day to one per hour in less than four months.
That deserves more weight than the old argument that Figure had little more than impressive demos.
Its robots are also back inside BMW's Spartanburg plant. Figure 03 is working on a more complicated sequencing task that involves picking differently oriented parts, moving carts and adjusting its body while handling objects.
Figure has also signed a commercial agreement with Catalyst Brands for logistics deployments.
Even so, the production gap remains large. AgiBot's previously discussed milestones imply roughly 55 additional robots per calendar day during its latest two production ramps. Figure's stated one-per-hour rate would equal 24 robots a day even if that line ran continuously around the clock.
The definitions and factory schedules are obviously different, so we should not treat that as an exact productivity comparison. It does show that Figure is moving into the same order-of-magnitude conversation while still operating below AgiBot's demonstrated output.
Figure's freshest expansion is happening on the AI side. Its Index data-collection network has attracted more than 44,000 weekly active contributors and accumulated more than 16 million uploaded videos. The company then signed an infrastructure agreement with Nscale covering access to as many as 100,000 Nvidia Vera Rubin GPUs, with an initial compute commitment valued at $3.5 billion.
That is an enormous bet on future robot intelligence.
Apptronik is taking a somewhat different route.
It has now raised more than $935 million in its Series A, with TechCrunch reporting a valuation around $5.3 billion. Mercedes-Benz, Google, Jabil, GXO and John Deere are among the companies around its ecosystem.
Its newest Apollo 2 robots are running across Apptronik's Robot Park network and customer sites, collecting training data with Google DeepMind. Jabil is also supposed to help manufacture Apollo while testing the robots inside its own operations.
The missing piece for Apptronik remains the same: we do not have a large published fleet count or revenue number.
Figure is therefore ahead of Apptronik on visible manufacturing scale today. Both still trail AgiBot on robots actually produced and shipped.
This chart, featured in our robotics market deck, compares the main business model options for warehouse AMR robotics providers
Is Agility Robotics further ahead than the shipment rankings suggest?
Agility Robotics is much further ahead on proven repetitive work than its relatively small fleet would suggest.
Digit's strongest evidence comes from GXO.
The two companies moved from a pilot into a multi-year Robots-as-a-Service agreement, putting Digit into live warehouse operations rather than keeping the robot inside a demonstration lab.
Agility says Digit has now moved more than 100,000 totes at GXO's Flowery Branch facility.
That is the kind of repetition commercial robotics needs to prove. Moving one tote in a demo says very little. Repeating the workflow one hundred thousand times inside a customer's operation begins to tell us something about reliability, integration and whether the process can survive normal warehouse conditions.
The fleet itself is still small. Independent deployment trackers can verify only a fraction of the thousands of robots being shipped by Chinese manufacturers, while Agility itself has referred to dozens rather than thousands of Digit units in the field.
So Agility is not the fastest-growing robotics company by unit volume.
It may be ahead on another question: how much useful work has each deployed humanoid proven it can do?
That distinction is worth keeping. One day, ten robots that each perform a paid warehouse task reliably may tell us more about the future than one thousand robots sold mainly for research.
AgiBot wins the growth comparison today. Agility has one of the cleaner proofs that humanoid robots can earn their way into a real workflow.
Could FieldAI, Bedrock or Locus be growing faster outside humanoids?
FieldAI is the strongest non-humanoid challenger we found, but its public numbers still fall short of AgiBot's current commercial ramp.
Business Insider reported that FieldAI had surpassed $100 million in combined revenue and customer contracts across more than 30 customers.
The wording is important. Revenue already recognized and the total value of signed contracts are two different things, so we cannot compare that $100 million directly with AgiBot's reported revenue.
Still, FieldAI's customer expansion looks substantial.
Its software runs on different robot platforms rather than requiring customers to buy one proprietary machine. Boston Dynamics is working with FieldAI on autonomous Spot deployments in construction. Customers cited in the companies' case studies say some pilots have expanded into broader fleet deployments.
The newest development is bigger. Caterpillar is now collaborating with FieldAI on physical AI and autonomous industrial equipment. A startup whose software can spread across Caterpillar machines has a very different scaling path from a startup that must manufacture every robot itself.
Bedrock Robotics is younger and smaller commercially, but it has made one of the sharpest recent moves from testing into real work. Its autonomous excavators are now operating without anyone in the cab on three major construction projects involving Sundt Construction, Champion Site Prep and Zachry Construction.
Bedrock has raised more than $350 million and reached a $1.75 billion valuation. What we do not have yet is a useful revenue number, fleet count or cost-per-cubic-yard comparison.
Locus Robotics sits at the opposite end of the maturity spectrum.
Sacra estimates Locus reached about $180 million of annual recurring revenue by mid-2026, up from $165 million at the end of 2025. Its robots have completed more than seven billion warehouse picks, with the latest billion taking only about four and a half months.
Locus is already a substantial robotics business. Its percentage revenue growth simply looks much more normal than AgiBot's.
The comparison shows why humanoid headlines can distort the broader robotics market. FieldAI may have the easiest software scaling model. Bedrock is attacking an enormous existing equipment base. Locus already has recurring revenue and billions of completed tasks.
None currently combines the same revenue acceleration, manufacturing growth and market-share gain that puts AgiBot first.
| Company | Best current growth evidence | What stops it taking first place |
|---|---|---|
| FieldAI | $100M+ in reported revenue and contracts; 30+ customers; Caterpillar partnership | Revenue and contract value are combined |
| Bedrock Robotics | Fully autonomous excavators on three live infrastructure projects | Fleet and revenue still undisclosed |
| Locus Robotics | ~$180M estimated ARR; 7B+ warehouse picks | Revenue growth is much slower |
| AgiBot | Multi-fold revenue growth plus world-leading humanoid shipments | Revenue quality and deployment mix remain less transparent |
If you want more recent data on this point, please see our latest robotics market report.
This chart, featured in our robotics market deck, breaks down revenue across customer segments in the robotics market
Can AgiBot keep growing this fast?
AgiBot can keep growing very quickly, but another 17-fold annual revenue increase is neither necessary nor realistic from its new base.
Once a hardware company reaches billions of renminbi in sales, basic arithmetic changes the story.
Repeating 2025's growth rate would push AgiBot toward RMB18 billion of annual revenue almost immediately. Repeating it again would create an absurdly large robotics company within another year.
A more believable path is the one the company is already describing: several-fold growth rather than another order-of-magnitude jump.
The reported RMB4 billion 2026 target would be roughly four times the 2025 figure. The company's public goal of more than RMB10 billion in 2027 would require another increase of around 2.5 times.
Those are still huge growth rates for a company manufacturing physical machines.
The wider market can support more volume for a while. Smart Analytics Global expects worldwide humanoid shipments to approach 60,000 units for 2026, while Counterpoint expects more than 50,000. Both expect industrial and service deployments to become a larger part of the mix.
The harder problem will eventually be demand quality.
A university may buy a research robot once. An exhibition company may need a limited fleet. A data-collection center can absorb robots while embodied-AI investment is booming.
A factory with a robot that reliably saves money behaves differently. It can order another hundred, roll the system into another plant and keep expanding for years.
AgiBot's next growth phase depends on creating more of that second kind of customer.
What could knock AgiBot off the top spot?
AgiBot will lose the fastest-growth title if its extraordinary shipment numbers fail to turn into repeatable, profitable demand while one of the software-first companies starts scaling through other manufacturers' installed fleets.
The first risk is the quality of AgiBot's revenue.
As seen above, AgiBot generated RMB1.05 billion in 2025, but we still do not have public information on gross margin, customer concentration, the share of sales connected to government projects, or how much business comes from investors and ecosystem partners.
Deng Taihua has openly said AgiBot is not rushing to achieve positive net profit because it wants to keep investing.
That is reasonable for a three-year-old company growing this quickly. It also means we cannot yet tell how expensive the growth is.
The second risk comes from software.
If Skild AI can put the same intelligence layer across ABB, Universal Robots, Fetch systems and other hardware, each additional deployment may require far less capital than AgiBot needs to manufacture another machine.
FieldAI has a similar opportunity with Caterpillar, Boston Dynamics and other platforms.
The third risk is Figure.
Figure's hardware scale still trails AgiBot, but a production line making one Figure 03 per hour, a growing BMW deployment, an enormous data-collection operation and billions of dollars committed to future compute make it much harder to dismiss than it was a year ago.
Finally, Unitree has already shown that high-volume robotics can be profitable. If AgiBot reaches several billion renminbi in revenue but remains structurally loss-making while customers refuse to expand early pilots, the growth will look much less impressive.
So AgiBot's lead is real, but it is a lead in a race whose important metric is changing quickly—from building robots, to selling robots, and eventually to proving that customers want many more of them after seeing what the first ones can actually do.
This chart, featured in our robotics market deck, shows how home cleaning robot technology has evolved over time
Which robotics startup is growing the fastest?
AgiBot is currently the fastest-growing major robotics startup we can identify from public commercial evidence.
The case is stronger now than it was even a few months ago.
AgiBot has moved from a tiny 2023 business to more than RMB1 billion of annual revenue in less than three years. Recent reporting suggests a single quarter may already be approaching the previous full year's sales.
Its manufacturing operation has reached five-figure cumulative production while increasing output several times over.
Two independent market-research firms currently rank AgiBot first worldwide in humanoid shipments, and one estimates its shipments are growing more than twice as fast as the already booming global market.
The challengers each win on a narrower definition.
Skild AI probably has the most interesting software-growth case. FieldAI is already building a serious cross-platform industrial business and has just added Caterpillar to a growing list of major partners. Figure AI has made the biggest recent improvement in US humanoid production and is spending at a scale few robotics startups can match. Agility Robotics has unusually good evidence that a humanoid can perform the same paid warehouse workflow tens of thousands of times. Locus Robotics remains a much more mature recurring-revenue company.
None gives us the same combination of current revenue acceleration, manufacturing growth and rising shipment share.
There is still an important caveat. AgiBot's growth quality is harder to judge than its growth speed. We know much less about margins, repeat purchases and the economic value of each deployment than we know about how many robots are leaving its factories.
That is enough to keep the long-term winner uncertain.
For the question we are actually asking, though, we do not need to predict the long-term winner. We need to identify which robotics startup is expanding fastest today.
For now, that company is AgiBot.
If you want more recent data on this point, please see our latest robotics market report.
OUR METHODOLOGY
The question “Which robotics startup is growing the fastest?” is less obvious than it looks because robotics companies can expand through very different mechanisms. We therefore broke the question into commercial growth, shipments and production, paid deployment, customer expansion and repeat use, then used funding and valuation only where they helped explain the pace or direction of growth.
We gave the most weight to activity already happening: revenue generated, robots produced or shipped, deployments operating with customers, and fleets expanding after initial trials. Announced factory capacity, financing and valuation were treated as evidence of future growth capacity rather than proof that growth had already occurred.
We also kept different types of evidence separate. Revenue, contract value, production, shipments, completed tasks and funding are not interchangeable, even when the headline numbers are large. Where two market-research firms used different definitions, we focused on conclusions that survived both methodologies rather than forcing one exact estimate.
The final ranking came from the aggregation of those dimensions. A company did not need to lead every measure, but it needed several recent commercial indicators to point in the same direction. That is why the comparison includes humanoid manufacturers such as AgiBot and Figure, software-first companies such as Skild AI and FieldAI, and more mature warehouse-robotics operators such as Locus Robotics.
For AgiBot, the core evidence comes from independent humanoid-shipment estimates and direct production milestones. Key sources include Smart Analytics Global's H1 2026 humanoid shipment analysis, Counterpoint Research's independent H1 2026 shipment analysis, and AgiBot's disclosure on its 15,000th robot and production milestones.
For the Unitree comparison, we prioritized public-market information rather than secondary estimates, including Shanghai Stock Exchange material on Unitree's IPO review and 2025 financial results and the Shanghai Stock Exchange's later Unitree listing update.
For the software-first and US humanoid challengers, key sources include Skild AI's Series C disclosure, Skild's ABB and Universal Robots deployment announcement, NVIDIA's confirmation of Skild-related physical-AI deployments, Skild's acquisition of Zebra Technologies' robotics business, Figure's Figure 03 manufacturing-ramp disclosure, Figure's BMW deployment update, and Figure's Nscale compute agreement.
For evidence of real-world work beyond raw shipment counts, we used customer and partner-side sources where possible, including GXO's commercial agreement with Agility Robotics, Agility Robotics on Digit moving more than 100,000 totes, Boston Dynamics on its FieldAI partnership, Locus Robotics on surpassing seven billion warehouse picks, and Bedrock Robotics' announcement of fully autonomous excavator deployments on live US infrastructure projects.
This methodology is designed to answer a narrow question: which major robotics startup is expanding fastest based on the freshest observable commercial evidence. It is not a prediction of which company will ultimately dominate robotics.
In our robotics market deck, we identify pain points entrepreneurs should prioritize
Related blog posts
