Low-cost humanoids: which startup is ahead?

Which startups belong in the low-cost humanoid race?

The core low-cost humanoid race currently has four serious startups: Unitree, Booster Robotics, Noetix Robotics and 1X Technologies.

We count robots that have two legs, two arms, a published or directly offered price of about $20,000 or less, and some credible path to regular production. This keeps the comparison focused on products that a university, developer, school or early consumer could realistically order.

Unitree qualifies through the $4,900 R1 Air, the $5,900 standard R1 and the $13,500 G1. Booster enters with the compact K1, whose international starting price is around $5,999. Noetix has pushed the lower boundary much further with Bumi, launched at RMB9,998, or roughly $1,400. The $20,000 NEO places 1X at the upper edge of our definition.

EngineAI remains useful as a technical benchmark, although we exclude it from the final ranking. Its RMB42,000 SA01 has no arms, while the complete PM01 now costs RMB188,000, roughly $26,000. Comparing either product directly with a complete sub-$20,000 humanoid would distort the result.

Funding totals also require care. Chinese robotics companies often announce individual rounds without publishing a complete cumulative figure. We therefore use confirmed totals where available and lower bounds elsewhere rather than pretending that every number is equally precise.

Startup	Main low-cost humanoid	Primary market	Starting price	Reported cumulative funding
Unitree	R1 and G1	Developers, universities and research labs	$4,900	About $252 million before its IPO
Booster Robotics	K1	Education, research and robot competitions	About $5,999	More than RMB1 billion; its latest round alone approached that amount
Noetix Robotics	Bumi	Families, schools and beginner developers	RMB9,998, about $1,400	More than RMB1 billion; nine financing rounds have been disclosed
1X Technologies	NEO	Household assistance	$20,000 or $499 per month	About $126 million
EngineAI, adjacent benchmark	SA01 and PM01	Locomotion research and industrial development	RMB42,000 for SA01; RMB188,000 for PM01	More than RMB1 billion, with later rounds adding an undisclosed amount

Is Unitree already leading low-cost humanoids?

Among low-cost humanoid startups, Unitree is currently ahead by a wide margin as a hardware business.

Unitree’s latest IPO prospectus gives us far more evidence than most private competitors provide. The company shipped around 5,500 humanoids in 2025, generated approximately RMB1.7 billion in total revenue and reported about RMB600 million in adjusted profit. Its gross margin reached 59.8%, which is unusually high for a young hardware manufacturer still expanding production.

China’s securities regulator has now approved Unitree’s IPO registration on the Shanghai STAR Market. That approval adds weight to the disclosed figures because the company has had to expose its finances, customer mix and operating risks to regulatory review. Unitree is seeking roughly RMB4.2 billion to fund new robots, manufacturing capacity and embodied-AI models.

Booster is the closest operational challenger, but the distance remains large. Booster has disclosed more than 1,000 cumulative K1 shipments, compared with Unitree’s several thousand humanoids in one year. Noetix has strong sales and order announcements but no comparable delivered fleet. 1X is still preparing its first broad wave of NEO customer deliveries.

Unitree’s lead has a clear limit. During the first nine months of 2025, 74% of its humanoid revenue came from research and education, while industrial customers contributed only 9%. More than half of that industrial portion involved tours and displays inside commercial spaces. Unitree has already built the strongest affordable humanoid hardware company, but reliable robotic labor remains largely unproven across the market.

If you want more recent data on this point, please see our latest humanoid robotics market report.

Who has actually delivered low-cost humanoids to customers?

On actual low-cost humanoid deliveries, Unitree comes first and Booster is the only startup with a visible four-figure fleet behind it.

Unitree’s roughly 5,500 humanoid shipments give it a scale that most competitors still describe as a future target. The figure covers completed shipments rather than reservations or factory capacity. It also made Unitree one of the two companies responsible for most of the 13,000-plus humanoids shipped globally during 2025, according to Omdia figures reported by the Associated Press.

Booster has shipped more than 1,000 K1 robots to over 400 customers across more than 20 countries. That customer count suggests an average of at least 2.5 robots per client. Booster has not published the distribution, so the average is only directional. Even so, it points to a broader installed base than a collection of isolated showcase units.

Noetix has disclosed compelling sales numbers without giving us an equally clear delivery count. Its first batch of 500 Bumi robots reportedly sold out within two days, and the company later said several thousand units had been sold through online and offline channels. Noetix also announced a separate 1,000-unit agreement with Huicheng. “Sold,” “ordered” and “delivered” can describe different stages, so we give these figures less weight than completed shipment totals.

1X has opened NEO orders and begun producing robots at its California factory. Its official schedule still describes customer deliveries as starting during 2026, and no independently confirmed installed base has emerged yet. 1X has crossed into production, although it has not crossed the harder threshold of proving repeatable customer delivery.

Which low-cost humanoid is cheapest, and which offers the best value?

Noetix Bumi is currently the cheapest complete humanoid, while Unitree R1 offers the strongest value for serious robotics development.

Bumi’s launch price of RMB9,998 comes to roughly 28% of the price of a Unitree R1 Air and 23% of a Booster K1. Noetix reaches that price with a 98-centimetre, 17-kilogram body, 21 degrees of freedom and basic computing power of six TOPS. The robot is built for companionship, visual programming, education and preloaded movement rather than heavy manipulation.

Unitree’s R1 Air costs around three and a half times more, but the buyer receives a taller 123-centimetre platform, open joint and sensor interfaces, native simulation support and access to a broader development ecosystem. The base model has 20 degrees of freedom, while more expensive configurations add joints, computing and optional hands.

Booster K1 sits close to the R1 in price and serves a narrower research audience. It has 22 degrees of freedom, a stereo depth camera and configurations ranging from 48 to 200 TOPS. Universities studying locomotion, autonomous football or reinforcement learning may prefer Booster because its software and competition community are built around those tasks.

The Unitree G1 costs more but provides a larger research platform with lidar, depth sensing and optional dexterous hands. NEO is more expensive again because 1X is selling the promise of household help, including software, support and remote expert assistance. Its economics depend on repeated useful work inside homes, which no public deployment data currently proves.

Robot	Starting price	Size and joints	Best current use	Main limitation
Noetix Bumi	About $1,400	98 cm, 21 degrees of freedom	Education, companionship and beginner programming	Limited computing and simple hands
Unitree R1 Air	$4,900	123 cm, 20 degrees of freedom	Affordable humanoid development	Base version lacks dexterous hands
Booster K1	About $5,999	95 cm, 22 degrees of freedom	Research, teaching and robot sports	Small body and modest payload
Unitree G1	$13,500	About 132 cm, up to 43 degrees of freedom	Advanced research and manipulation experiments	Capable versions become much more expensive
1X NEO	$20,000 or $499 monthly	Full-size household design	Early home assistance	Deliveries and autonomous usefulness remain unproven

If you want more recent data on this point, please see our latest humanoid robotics market report.

Which startup has found real demand for low-cost humanoids?

Commercial demand for low-cost humanoids currently favors Unitree, with Booster showing the clearest repeatable niche behind it.

Unitree has already turned humanoids into its largest revenue category. Humanoids generated slightly more than half of company revenue in 2025, overtaking the quadruped robots that originally built Unitree’s business. Customers are buying a growing product family rather than one experimental release.

The customer mix remains heavily tilted toward research, education, entertainment and government-backed projects. These buyers still count. They pay for robots, expose hardware to real users and create a developer base. However, a university purchasing one G1 offers less recurring demand than a factory expanding from ten robots to several hundred.

Booster’s market is smaller but unusually coherent. More than 400 customers use its robots across education, academic research and competitions. The company also said its backlog had reached hundreds of millions of yuan after a sharp rise in orders. A focused customer group often gives a young hardware company a better foundation than unrelated pilots across ten industries.

Noetix has revealed something different: consumers will consider a humanoid when the price approaches that of premium electronics. The fast sale of Bumi’s initial batch, several thousand reported sales and the Huicheng agreement all point in the same direction. We still need cancellation, delivery and usage data before calling Bumi a consumer success.

1X has attracted attention and refundable deposits for NEO, but it has not disclosed the number of buyers, the conversion rate from deposits to full payment or recurring subscription demand. The home market could eventually dwarf the education market. Today, Unitree and Booster have much firmer evidence that customers will complete a purchase.

Which low-cost humanoid startup is growing fastest now?

Booster is growing fastest from a smaller base, while Unitree is adding more business in absolute terms.

Booster reported that first-quarter shipments rose 500% from the previous year. That quarter alone matched the company’s entire shipment volume during the first eight months of 2025. Orders signed during the first two months of the year increased by more than 800%, and Booster reached positive operating cash flow in December 2025.

These percentages make Booster the fastest-moving challenger, although its starting point was far below Unitree’s. Even several years of triple-digit growth would be required to close the gap if Unitree also keeps expanding.

Unitree’s revenue increased 335% during 2025, adding more than RMB1 billion in annual sales. Founder Wang Xingxing has discussed shipping between 10,000 and 20,000 humanoids during 2026. That goal would mean roughly doubling to nearly quadrupling the previous annual volume. We treat it as a target until the company reports completed shipments.

Noetix is growing quickly in the ultra-cheap segment. During China’s recent 618 shopping festival, one industry report said Bumi’s gross merchandise value rose 254% and unit sales increased 160%. Noetix has also rolled out new app features and over-the-air software updates, which shows that the company is supporting Bumi after launch rather than moving immediately to another prototype.

1X has moved from product development into factory production. That is an important step, but factory output only becomes commercial growth when robots reach paying households. For now, Booster has the strongest measured acceleration, Unitree has the largest absolute expansion, and Noetix has the most interesting consumer-price momentum.

Which low-cost humanoids are genuinely ready to use?

Product maturity currently favors Unitree R1, Unitree G1 and Booster K1 because customers can buy them, develop on them and find other users running the same hardware.

Unitree publishes prices, detailed specifications, software interfaces and simulation support for both R1 and G1. Thousands of shipped robots mean recurring issues with joints, batteries, control software and repairs have had more opportunities to surface than they would in a fleet of dozens.

Booster has reached a similar level within a narrower market. K1 is shipping internationally, its hardware has multiple computing configurations, and outside research teams actively write software for it. Booster has also built documentation and open-source tools around locomotion rather than leaving every customer to create a training pipeline alone.

Bumi has entered mass production and reached real buyers. Its product page includes multiple configurations, battery information, programming features and user documentation. Noetix is therefore beyond the prototype stage. Its current use cases, however, remain closer to education and interactive entertainment than flexible physical work.

NEO is the least mature product in the core group from a customer-deployment perspective. 1X has a functioning design, an operational factory and a public ordering process, but the first owners are still waiting for broad delivery. NEO could advance quickly once households begin publishing months of real usage.

Startup	Current product stage	Best evidence of maturity	What remains unproven
Unitree	Regular commercial shipment	Large shipped fleet, public store, SDKs and multiple product generations	Long-term industrial productivity
Booster Robotics	Regular commercial shipment	International customers, open-source tools and competition use	Expansion beyond research and sports
Noetix Robotics	Early mass-market shipment	Initial batches sold, product documentation and ongoing software updates	Retention, reliability and useful consumer behavior
1X Technologies	Factory production and early access	Operational factory and orderable product	Confirmed household deliveries and repeated autonomous chores

Can any low-cost humanoid work autonomously today?

General autonomy remains absent from low-cost humanoids today, even when the hardware can walk, talk and perform impressive routines.

1X describes NEO as arriving with basic autonomy through its Redwood model. When NEO encounters a task it does not know, a human expert can remotely supervise the robot at a scheduled time. That system could help early users complete more chores and generate valuable training data. Still, early household performance partly depends on human labor behind the screen.

Unitree sells flexible bodies for developers to program. An R1 or G1 can run trained policies, follow prepared sequences and support outside perception systems. The buyer still has to build or integrate much of the intelligence required for a useful task. Unitree’s own product pages warn buyers that some functions remain under development and that the humanoid industry is in an early stage.

Booster robots can play fully autonomous football when university teams add their own perception, strategy and control software. That demonstrates real autonomous movement in a structured, adversarial setting. It does not mean a standard K1 arrives ready to tidy a room or unload a truck. The strongest intelligence often belongs to the research team running the robot rather than Booster’s standard product.

Bumi supports voice interaction, visual programming, object recognition and preloaded movements. With six TOPS in the base system and simple end effectors, its practical role is closer to an interactive programmable device. Noetix may improve Bumi through software updates, but the current hardware was designed around affordability.

The low-cost humanoid market has made bodies much cheaper faster than it has made autonomy reliable. Buyers can now purchase a walking platform for the price of a used car or, in Bumi’s case, a premium laptop. They still need software, supervision and a carefully limited task.

If you want more recent data on this point, please see our latest humanoid robotics market report.

Which startup has the best-performing humanoid hardware?

Raw physical performance currently favors EngineAI, while Unitree offers the strongest all-round low-cost platform and Booster has the best competition record.

EngineAI’s PM01 advertises speeds above two metres per second, joint torque of up to 145 newton-metres and close to two hours of battery life. It has demonstrated running, front flips and human-like walking. The current RMB188,000 price keeps PM01 outside our core low-cost ranking, yet its actuators and dynamic control show what nearby competitors can achieve.

Unitree balances speed, stability, sensing and extensibility across several body sizes. The G1 is used in a growing number of outside research projects, which tells us more than company videos alone. One recent study built a grasping pipeline on a real G1 and reduced the preparation time for a new object from one or two days to around 30 minutes. The researchers reported successful grasps at five positions and later tested the approach on an automotive glue-application task. The intelligence came from the research system, but Unitree supplied hardware flexible enough to support it.

Booster’s evidence comes from repeated competition under shared rules. At the latest RoboCup, teams using Booster robots won the large, middle and small humanoid divisions. The official results show Booster hardware beneath most of the leading teams, while each team supplied its own autonomous software. It is a hard, useful test of stability, recoverability and developer access.

Noetix has achieved respectable movement at a much lower cost, but Bumi’s small body and simple hands reduce its usefulness for manipulation. NEO focuses on quiet, safe movement around people. That design may eventually matter more in homes than running speed, although customer evidence is still missing.

Which startup has the strongest developer ecosystem?

Unitree currently has the broadest low-cost humanoid developer ecosystem, while Booster has built the most committed specialist community.

Unitree benefits from volume. More shipped robots create more tutorials, research papers, spare-part knowledge, software experiments and engineers who already understand the platform. The company also sells several humanoids and quadrupeds that share parts of its control and development stack. A laboratory can start with an R1, move to a G1 or combine humanoids with Unitree robot dogs without changing suppliers.

The company’s open joint and sensor interfaces help outside teams add their own perception, reinforcement-learning or manipulation systems. Unitree’s latest IPO documents also show that nearly half of the planned model-development funding will go toward the software intelligence behind its robots. This is where Unitree is currently weaker than its hardware reputation suggests, so the investment has a clear purpose.

Booster’s smaller ecosystem is easier to see in action. Its Booster Gym framework covers reinforcement-learning training, simulation and transfer to a physical T1. The newer Booster Lab project adds motion-data preparation and cross-platform testing on T1 and K1. These tools reduce the amount of infrastructure each research group needs to rebuild.

Booster robots also powered most teams in the latest RoboCup humanoid competition and every divisional winner. That concentration gives Booster a valuable loop: researchers improve the software, competitions expose hardware problems, and the resulting tools make the platform more attractive to the next group.

Noetix is building a more accessible ecosystem through visual programming, educational models and mobile controls. 1X is taking the opposite path. It controls NEO’s body, AI, factory and remote-support service as one closed consumer product. That model could create a strong proprietary dataset, although it will probably attract fewer independent developers.

If you want more recent data on this point, please see our latest humanoid robotics market report.

Which startup can really manufacture low-cost humanoids at scale?

Unitree is currently the only low-cost humanoid startup that has proved high-volume production alongside strong financial margins.

The company manufactures many core components internally, including motors and control systems. That vertical integration helped Unitree lower prices while keeping a gross margin close to 60%. Its latest accounts also show positive adjusted profit rather than a business financed entirely through venture capital.

Unitree’s private funding of roughly $252 million is similar to the amount raised or disclosed by several rivals. The difference appears in output. Unitree has already converted that capital into several product generations, meaningful revenue, a profitable operation and an approved public listing. We therefore rank it first on funding efficiency as well as manufacturing.

Booster has made the next strongest case. Its K1 fleet has crossed four figures, shipments are rising rapidly and the company has reported positive operating cash flow. The latest financing of nearly RMB1 billion should help it expand production and overseas service. Its challenge is keeping quality stable while moving beyond a compact research platform.

Noetix says Bumi is the first bipedal humanoid mass-produced in China below RMB10,000. Selling hardware at that price requires aggressive component sourcing and a simple product architecture. The early sales show that production exists, but Noetix has not disclosed factory output, failure rates or gross margins.

1X has built the most ambitious new factory among the challengers. Its 58,000-square-foot Hayward site has stated capacity for up to 10,000 NEO robots per year, with a path toward more than 100,000 annually by the end of 2027. The factory is currently producing robots for internal testing and upcoming customers. Capacity describes what a plant could build under the right conditions; completed deliveries will show how much it can build reliably.

Does each low-cost humanoid startup lead a different market?

Each leading low-cost humanoid startup currently owns a distinct position, although Unitree covers the widest part of the market.

Unitree leads affordable general-purpose development hardware. Its range stretches from the lightweight R1 to the more capable G1, giving it access to individual developers, universities, corporate labs, entertainment buyers and early industrial projects.

Booster leads compact humanoids for research, education and robot sports. It is a narrow market, sure, but a real one with shared benchmarks, recurring competitions and customers who actively modify the product. Booster’s focus has helped it build deeper community adoption than a broad collection of unrelated pilots would have produced.

Noetix leads the ultra-cheap consumer and education category. Bumi asks a different question from G1 or NEO: can a humanoid become affordable enough to buy for learning, entertainment or companionship even before it can perform valuable labor? Noetix currently has the clearest answer.

1X leads the home-humanoid bet. NEO’s soft exterior, tendon-driven body, quiet movement and remote-support model are designed around houses rather than laboratories. 1X could eventually create the most valuable business in this comparison because household demand is enormous. That upside remains tied to future performance rather than current deployments.

EngineAI leads dynamic locomotion close to the low-cost boundary. Its strongest complete robot now exceeds our price limit, so it competes more directly with higher-end research platforms than with Bumi or R1 Air.

What could change the low-cost humanoid ranking?

The low-cost humanoid ranking will change only when a challenger proves a milestone that Unitree has not already reached.

Booster needs to sustain its current growth while finding customers beyond football, research and education. A few thousand annual deliveries into logistics, inspection or commercial service would narrow the gap quickly because Booster already has credible hardware and a strong developer community.

Noetix needs to show that Bumi buyers keep using the robot after the novelty wears off. Delivered-unit totals, active monthly users, software engagement, warranty claims and repeat institutional orders would reveal whether the sub-$1,500 category has durable demand.

1X faces the clearest test. NEO must reach a meaningful number of homes, complete chores repeatedly and reduce its reliance on remote experts. A fleet of several thousand frequently used household robots would move 1X toward the top of the ranking even if its hardware volume remained below Unitree’s.

Unitree’s risk comes from the quality of its existing demand. Research laboratories, schools and entertainment buyers helped build a large business, but these customers may order in small batches and replace robots slowly. Unitree must turn its hardware advantage into industrial workflows or recurring services before cheaper rivals reproduce its manufacturing strengths.

The broader market could also shift toward a use case where humanoid shape provides little value. Fixed robotic arms, wheeled robots and specialized machines frequently complete tasks more cheaply and reliably. Low prices make humanoids easier to test, but customers will continue comparing them with every other form of automation.

Which low-cost humanoid startups are actually ahead?

Unitree finishes first today, Booster is the only operational challenger within sight, Noetix leads the price disruption, and 1X remains the highest-upside but least-proven bet.

Unitree deserves first place because it leads on the evidence that is hardest to fake: completed shipments, recognized revenue, profit, product availability and years of manufacturing experience. Its lead is much stronger in affordable hardware than in autonomous work, but hardware remains the foundation every competitor needs.

Booster earns second place through real shipments, hundreds of customers, rapid growth, open-source development tools and dominant adoption in humanoid robot competitions. The gap with Unitree remains substantial, yet Booster has already progressed beyond the stage where one large announcement can create the appearance of leadership.

Noetix ranks third. Bumi has changed the price discussion and created early demand at a level other companies had not reached. Its ranking could rise quickly once Noetix publishes clearer delivery, usage and reliability data.

1X ranks fourth for now. NEO is the most distinctive product in the group, and 1X has built serious manufacturing infrastructure around it. The company is asking customers to pay $20,000, or nearly $6,000 per year through the subscription, before public evidence shows that NEO can handle enough chores independently. Ambition carries less weight than completed household use.

EngineAI sits outside the core ranking because its inexpensive SA01 lacks arms and its complete PM01 now exceeds the price ceiling. Within the broader affordable-humanoid market, EngineAI remains one of the strongest locomotion companies and could re-enter the ranking with a cheaper complete robot.

Rank	Startup	Why it ranks here now	What could move it
1	Unitree	Clear leader in shipments, revenue, profitability, product range and manufacturing experience	Proving repeated industrial work would make the lead much harder to challenge
2	Booster Robotics	Strong delivered fleet, fast growth, research adoption and the deepest robot-sports ecosystem	Expansion into commercial tasks and several thousand annual shipments
3	Noetix Robotics	Lowest complete-humanoid price and convincing early consumer demand	Transparent delivery, active-use and reliability figures
4	1X Technologies	Strongest home strategy, differentiated design and substantial factory capacity	Thousands of household deployments with limited remote supervision

If you want more recent data on this point, please see our latest humanoid robotics market report.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.