Humanoid Robot Market Size: $3B Today to $5T by 2050

What is the humanoid robot market size?

The direct answer: seven publisher-owned public report summaries reviewed on July 23, 2026 estimate the global humanoid robot market at $425 million to $7.8 billion in 2025. The median of the six sources with current 2025 base years is $2.66 billion, roughly $3 billion. That roughly 18-fold range is not a consensus: it shows how much the answer changes when a report counts only assembled robots versus hardware, software and services.

For 2030, BCC Research's $11 billion is now the only current public forecast with a matching 2025-to-2030 period; MarketsandMarkets' July 2026 revision moved its endpoint to 2035. Other publishers use endpoints from 2032 to 2036. Above all of that sit the bank scenarios that dominate headlines: Goldman Sachs' $38 billion total addressable market by 2035, and the 2050 projections from Morgan Stanley, Citi and UBS that run from $1.4 trillion to $7 trillion. Those are long-range scenarios, not revenue forecasts, and this page keeps them in a separate table for exactly that reason.

There is no official or audited global total. Every number below is a publisher's estimate. Market revenue, shipment volume, installed base and total addressable market answer different questions, so we keep the year, definition and geography attached to each figure. And because forecasts float free of the ground truth, we also show what the companies that disclose revenue actually earned in 2025: their sum, counting Unitree's entire company, was under $1 billion.

• 2025 estimate range: $425M to $7.8B across the public revenue forecasts; median of the six current 2025 bases: $2.66B.
• 2030 forecast: $11B (BCC Research); MarketsandMarkets' July 2026 revision now runs 2026 to 2035.
• 2035 scenarios: $15B (Interact Analysis) to $139B (Macquarie), with Goldman Sachs at $38B.
• 2050 bank scenarios: $1.4T (UBS) to $7T (Citi), with Morgan Stanley's much-quoted figure at $4.7T.
• Published revenue CAGRs: 25.8% to 50.6%, each tied to a different period and model.

Humanoid robot market forecast comparison

The tables use the values and scope disclosed on each publisher's own public page. They are estimates, not audited industry totals. On a phone, swipe horizontally inside each table to see every column.

MarketsandMarkets replaced its model in July 2026 with a 2026 base year, which leaves six sources with 2025 estimates. Their median is $2.66 billion, roughly $3 billion, and close to the $2.91 billion the older seven-source set produced. We show the median as a descriptive midpoint, not as a new market forecast. An average would imply a common measurement basis that these reports do not have.

Until July 2026 the cleanest same-period comparison was BCC Research versus MarketsandMarkets, whose 2030 estimates differed by $4.26 billion; the new MarketsandMarkets edition dissolved that pairing by moving to a 2026-to-2035 window. The broader table is useful for showing the range of published opinion, not for pretending every row measures the same market. It is also a live lesson in edition churn: a forecast you cite can be replaced under the same URL within a year.

The trillion-dollar forecasts: what the banks say about 2050

The numbers that dominate headlines and AI-generated answers do not come from the research houses above. They come from bank scenario models with 2050 horizons, and they disagree with each other by 5 times on dollars and more than 3 times on robots in use by 2050, before Bank of America's 3 billion arrives a decade later. That spread deserves its own table.

Read the table horizontally and the disagreement is the story. For the same year, 2035, Macquarie's figure is 3.7 times Goldman's. For 2050, UBS and Citi differ by 5 times in dollars, and the 2050 installed-base assumptions run from UBS' 300 million robots to Morgan Stanley's roughly 1 billion, with Bank of America's 3 billion dated 2060, a decade later. Morgan Stanley's much-quoted "$5 trillion" is actually $4.7 trillion in core humanoid revenue, with the rounder number including supply-chain and support services; its own report expects adoption to stay relatively slow until the mid 2030s before accelerating. Citi's figure is usually reported as 648 million robots, but Citi's own page says "almost 650 million," which is the honest level of precision.

ARK Invest circulates an even larger figure, roughly $24 to $26 trillion, but it belongs in a different category: it is a ceiling estimate of the revenue opportunity "at scale," with no target year, built partly by converting unpaid household labor into paid GDP. It is not a dated forecast and should not be compared against any number in the table.

Two dating notes matter when you see these figures quoted. Goldman's $38 billion is the oldest number still in wide circulation: it was published in February 2024 and has not been publicly revised, so citing it as a current view is a stretch the bank itself does not make. And the freshest full bank model is Bank of America's March 2026 update, whose near-term line, 90,000 shipments in 2026, can be checked against reality within eighteen months. That is a feature, not a weakness: it is the only mega-forecast that puts a testable number this close.

The reality check: what the industry actually earned

Every forecast above floats free of a small, verifiable base. In 2025, the companies that disclose relevant revenue at all, counting Unitree's whole company, quadrupeds included, summed to well under $1 billion. These are selected disclosed indicators with different scopes, not an industry total, and they are not additive.

Unitree, whose IPO was approved in early July 2026, reported 2025 whole-company revenue of RMB 1.69 billion, about $235 million at 2025 average exchange rates of roughly RMB 7.2 to the dollar, up 335% year on year, with humanoids at 51.5% of revenue in the first nine months of 2025. It is one of the only profitable companies in the sector. UBTECH, the only listed peer that breaks out a humanoid segment, reported RMB 820.6 million, about $113 million, of 2025 humanoid revenue on 1,079 units delivered, up from RMB 35.6 million the year before. an AgiBot announcement citing Omdia counted roughly 13,000 humanoid robots shipped worldwide in all of 2025, up around 480% from 2024, and Interact Analysis expects annual shipments to stay below 100,000 until the early 2030s, with a $15 billion market by 2035, the most conservative current 2035 estimate.

One counting dispute is worth flagging because it shows how soft even the unit data is: Unitree says it shipped more than 5,500 humanoids in 2025, while Omdia credits it with about 4,200. Both can be defended depending on what counts as a humanoid and when a shipment is booked. If the two best-documented companies in the industry differ by 30% on a four-digit number, treat every third-decimal market figure accordingly.

The gap math is the point. From a base of roughly $1 to $3 billion in 2025, Goldman's $38 billion by 2035 requires sustained growth around 35 to 40% a year for a decade, aggressive but in line with 2025's actual unit growth. Morgan Stanley's $4.7 trillion by 2050 requires that growth to continue for another fifteen years after that. Neither is impossible; both are extrapolations from a base year in which the entire industry earned less than a single mid-sized robotics company's annual revenue.

Forecasts versus units: what the models assume

Revenue forecasts hide their most testable claim, the number of robots. Here is what the public models say in units, which is where they can be checked first.

Read each cell with its label: shipments are annual flows, installed base counts robots in operation, and the China-only figures sit inside the global ones. The 2026 row is the one to watch. Bank of America's 90,000 and the roughly 50,000 cluster from Morgan Stanley and Deutsche Bank are predictions about this calendar year, against a 2025 actual of about 13,000. By early 2027 one of those numbers will look prescient and the others will not, which will say a great deal about whose 2030 and 2050 models deserve attention.

How we selected and compared sources

We reviewed publisher-owned public pages available on July 23, 2026. To enter the main comparison, a page had to describe a global humanoid or humanoids market, disclose a 2025 US-dollar revenue estimate, state a future endpoint and belong to the organization publishing or distributing the named research. We excluded aggregator copies and narrower component-only submarkets from the headline range.

This is a comparison of public summaries, not an independent audit of the paid reports or their underlying models. Where the public page does not disclose a revenue component, segmentation rule or regional assumption, we say so. Bank scenario models, Goldman Sachs included, are kept outside the seven-source revenue table because they describe total addressable market scenarios rather than directly comparable report series; they get their own table above with the same source discipline.

What each forecast is actually measuring

Frost & Sullivan: assembled-robot revenue only

The Frost & Sullivan report distributed by Research and Markets gives the lowest 2025 estimate: $425 million, rising to $4.75 billion in 2032 at more than 41.2% CAGR. Its public scope is unusually specific. It counts revenue from fully assembled humanoid robots sold by original equipment manufacturers and excludes component-level revenue and downstream services. That narrower denominator helps explain why its starting value is below the other six.

BCC Research: commercially deployed humanoids

BCC Research estimates the market at $1.9 billion in 2025 and $11 billion in 2030, a 42.8% CAGR. Its public description focuses on commercially deployed humanoid robots and segments revenue by wheeled versus bipedal type, industry vertical and region. The commercial-deployment wording is important because prototypes and demonstrations do not necessarily create market revenue. Note that BCC's numbers also circulate as "news": the same figures are routinely recycled through press-release wires and finance portals with fresh datelines, which makes one estimate look like several independent confirmations.

Grand View Research: revenue and unit volume

Grand View Research puts the market at $2.4 billion in 2025, $4.2 billion in 2026 and $40.5 billion in 2033. The published CAGR is 38.2% for 2026 to 2033. Its report page separates hardware and software, biped and wheel-drive machines, and multiple industrial and service applications. It also publishes a unit-demand forecast, which is useful because revenue can rise through either more robots or higher revenue per system.

360iResearch: the original seven-source median

360iResearch estimates $2.91 billion in 2025, $3.64 billion in 2026 and $14.53 billion in 2032, with a 25.8% CAGR from 2026 to 2032. This was the median of the original seven-source set and remains the lowest published CAGR; after the July 2026 MarketsandMarkets revision, the median of the six current 2025 bases sits at $2.66 billion, just below this figure. Its public introduction describes humanoids as human-form systems combining mobility, perception, manipulation and increasingly natural-language interfaces, but it does not itemize the revenue components included in the total.

MarketsandMarkets: hardware, software and services

MarketsandMarkets replaced its model in July 2026: the new edition estimates $5.41 billion in 2026 and $50.27 billion by 2035, a 28.1% CAGR, superseding the April 2025 edition's $2.92 billion for 2025 and $15.26 billion for 2030. Same publisher, same URL, endpoint more than tripled and growth rate cut, all within fifteen months. Its scope includes biped and wheel-drive humanoids, plus hardware, software and services, across applications from manufacturing to caregiving, education and search and rescue.

Fortune Business Insights: the highest published CAGR

Fortune Business Insights estimates $4.89 billion in 2025, $6.24 billion in 2026 and $165.13 billion in 2034. Its published 50.6% CAGR for 2026 to 2034 is the highest in this comparison. The report covers biped and wheel-drive robots, hardware and software, and industrial, household and service applications. It identifies Asia-Pacific as 42.6% of the 2025 market. One caution on precision: at the time of review, the same public page stated the 2025 market as $4.89 billion in its summary and $4.49 billion in its own FAQ. Fortune's own regional splits reconstruct to roughly $4.49 billion, so we show its 2025 figure as a range. A report that disagrees with itself by $400 million on the same page is a useful reminder of how much weight two decimal places should carry.

Future Market Insights: the broadest disclosed bundle

Future Market Insights publishes the highest 2025 estimate at $7.8 billion, followed by $10.69 billion in 2026 and $248.9 billion in 2036, a 37.0% CAGR. Its public definition includes complete humanoid platforms, integrated AI and control software, sensor and actuator subsystems sold with the platform, and bundled maintenance and deployment services. It excludes non-humanoid robots, standalone AI software and non-autonomous telepresence devices.

Goldman Sachs: a 2035 total addressable market scenario

Goldman Sachs Research projected in February 2024 that the total addressable market could reach $38 billion by 2035. The public article links that scenario to a 1.4 million-unit shipment estimate, a 40% reduction in material costs, faster AI progress and a later expansion from industrial into consumer use. It does not publish a CAGR on the public page, so assigning one would require assumptions that Goldman Sachs does not state there.

Goldman Sachs also calls out the main reasons the scenario may not arrive on schedule: component bottlenecks, limits in manipulation and interaction, and the unproven viability of mass-produced general-purpose humanoids. The $38 billion figure is useful as a scenario, not as a settled market fact. In late 2025, Goldman's China channel checks, as reported by CNBC, added a nearer-term view: roughly 15,000 to 20,000 humanoids shipped in 2025, with orders and shipments that "could multiply" in 2026.

Why humanoid robot market forecasts disagree

A forecast is the output of a model. Change the boundary or one of the core assumptions and the result moves sharply. The main sources of disagreement are:

• Hardware versus full-stack revenue. A hardware-only estimate will be lower than one that adds software licences, integration, maintenance, training or robotics-as-a-service revenue.
• Commercial deployment versus development activity. Public summaries do not always make clear whether paid pilots, prototypes, research platforms and demonstration units count as market revenue.
• Biped-only versus broader humanoid definitions. Several reports include wheel-drive humanoids alongside bipeds, while readers often picture only two-legged general-purpose robots.
• Industrial versus consumer adoption. Factory and logistics deployments can be modelled from defined tasks. Home, elder-care and general consumer demand requires stronger assumptions about safety, reliability and price.
• Shipments versus installed base. A shipment forecast counts units sold during a period. Installed base counts machines still in operation. Orders, reservations, pilots and demos are different again.
• Average selling price. Revenue can rise even with similar unit forecasts when one model assumes more expensive hardware, software or support. Falling component costs can also increase units while reducing revenue per robot.
• Forecast horizon. A five-year forecast and a ten-year total addressable market scenario carry very different uncertainty. Comparing their endpoints without the year is misleading.

How to interpret a 26% to 51% CAGR

The seven published compound annual growth rates span 25.8% to 50.6%. Even the low end looks extraordinary because the starting market is small. CAGR describes the smooth annual rate needed to connect a model's start and end values. It does not mean the industry will grow by that exact percentage every year.

Early markets usually move unevenly. A large fleet order, delayed production ramp or revised definition can change one year sharply. High CAGR forecasts are therefore better read as a statement about the model's adoption curve than as a near-term sales guarantee.

China versus the United States

The most searched regional question has a two-part answer: China leads on units and manufacturing, the United States leads on capital.

Morgan Stanley estimated in June 2026 that China accounted for more than 80% of the roughly 16,000 humanoids deployed worldwide in 2025, and doubled its China shipment forecast for 2026 to 50,000 units, its second upgrade in six months, citing commercial verification, policy support and supply-chain feedback. It models China's humanoid market at $2 billion in 2026 growing to $15 billion by 2030, figures that roughly match Chinese state-affiliated institutes, which put the domestic market near RMB 20 billion by 2026 and RMB 100 billion by 2030. Beijing has backed the sector since a 2023 Ministry of Industry policy calling for a world-class humanoid supply chain by 2027, and the top two global shippers in 2025, AgiBot and Unitree, are both Chinese.

The American answer is capital: selected US startups command the largest disclosed private valuations in the sector. Figure reached a $39 billion private valuation on more than $1 billion of new funding in September 2025, Apptronik raised $520 million in February 2026 at a reported valuation above $5 billion, Agility Robotics announced a roughly $2.5 billion deal in June 2026 to become the first US-listed pure-play humanoid maker, and Tesla is converting a Fremont line for Optimus. Whether US capital or Chinese volume proves the better position is one of the genuinely open questions every 2050 scenario quietly answers one way or the other.

What IFR data can and cannot prove

The International Federation of Robotics is the strongest reference point here for broader robotics categories and deployment context. IFR separates industrial robots from service robots, then divides service robots into professional, consumer and medical categories.

The World Robotics 2025 service-robot executive summary reports activity in professional logistics robots and consumer domestic robots. The industrial-robot executive summary covers factory automation. Neither is a clean humanoid-market total: most industrial robots are arms, and most consumer service robots are not humanoids.

IFR data can test whether a forecast's adoption story is plausible in the wider robotics market. It cannot be added to a humanoid forecast or used as proof that the same number of humanoids has shipped.

Where adoption is most credible first

Near-term humanoid adoption is most credible in structured settings: factories, warehouses, logistics, automotive production, inspection, research and controlled service work. Tasks are easier to specify, workspaces can be engineered around the machine, and safety procedures are easier to enforce than in a normal home.

This is consistent with Goldman Sachs' 2030 base case, in which almost all projected shipments are industrial. It is also consistent with the application categories used by BCC Research, MarketsandMarkets and Grand View Research, and with Morgan Stanley's 2050 model, in which about 90% of its billion robots do industrial and commercial work rather than household chores.

Homes, hospitals, elder care and schools may become important later, but they demand higher reliability, privacy protection, certification and safe human interaction. The gap between a controlled pilot and an unsupervised deployment is one reason long-range consumer estimates carry more uncertainty.

What the spread means for investors

This page does not give investment advice, but the forecast spread has one investor-relevant implication worth stating neutrally: until 2026, there was almost no way to own pure-play humanoid exposure at all, and now there suddenly is. UBTECH listed in Hong Kong, Unitree's Shanghai IPO was approved in early July 2026 at an implied valuation around $5.9 billion, and Agility Robotics is heading to a US listing. If those listings complete, each will publish periodic public financials, audited at least annually, which will do more to discipline the forecast industry than any methodology note.

The valuation mathematics deserve the same scrutiny as the market forecasts. Figure's $39 billion private valuation is roughly 40 times a generous $1 billion estimate of the sector's entire disclosed 2025 revenue. That is not automatically wrong, venture prices are claims about 2035, not 2025, but it means the capital markets are already trading on the bank scenarios, not on the $3 billion market that exists today. When an analyst cites a trillion-dollar figure, check which table it came from before treating it as a valuation anchor.

Market size is not the same as robot price or availability

Market revenue does not tell a buyer what one robot costs, whether it is available, or whether it can perform a specific job. For purchase-level research, compare current humanoid robot prices, inspect which humanoid robots are for sale, and verify deployment evidence separately.

For current company activity, production plans and operating data rather than market forecasts, use the 2026 humanoid robot industry report. Our humanoid robot companies guide tracks the manufacturers behind those systems.

How to evaluate a humanoid market-size claim

Before quoting any market number, record the following details. If the public page does not disclose one, mark it as unknown instead of filling the gap with an assumption.

• Publication date, base year and forecast year.
• Revenue, shipment volume, installed base or total addressable market.
• Hardware, software, services and subscription treatment.
• Biped-only or biped plus wheel-drive definition.
• Included applications, especially research, pilots, industrial use and consumer use.
• Global and regional coverage, plus any countries singled out.
• Currency, nominal-versus-real treatment and average selling-price assumptions.
• Whether the number comes from a public summary, a paid report or an independently audited dataset.
• Whether the "news" article citing it is original reporting or a recycled press release with a fresh dateline.

Do not publish unsupported market-share rankings

Exact vendor shares require a public denominator, region, period and unit definition. A table claiming that one company owns 31% and another 27% can look authoritative while mixing orders, shipments, pilots, prototypes and installed systems. The Unitree case above is the live example: the company and the leading analyst firm differ by roughly 30% on its own 2025 shipments.

• Use company announcements as signals, not proof of global leadership.
• Separate firm orders from reservations, memoranda, pilots and demonstrations.
• Separate humanoids from industrial arms, autonomous mobile robots and consumer domestic robots.
• Prefer named scenarios and explicit caveats over one precise but undefined number.

Bottom line

Publisher-owned public report summaries put the 2025 global humanoid robot market between $425 million and $7.8 billion, with a median of $2.66 billion across the six current 2025 bases. BCC Research's $11 billion is the remaining current 2025-to-2030 forecast after MarketsandMarkets moved its window to 2026-2035. Above them, the bank scenarios run from Goldman's $38 billion by 2035 to Morgan Stanley's $4.7 trillion and Citi's $7 trillion by 2050, while the sector's disclosed company revenue in 2025 summed to under $1 billion on about 13,000 shipped robots.

The 18-fold spread at the base year and the 5-times spread at 2050 are the finding, not a nuisance to hide. These publishers describe different versions of the market, so the source, year, metric, geography and scope should always travel with the figure. Treat every forecast as a planning scenario, not as an audited count of deployed humanoid robots, and watch the 2026 shipment numbers: they are the first place the models meet reality.