The Humanoid Robot Industry Shipped 14,500 Units in 2025. Investors Valued It at $100 Billion.
The top 10 humanoid companies carry a combined valuation north of $120 billion on roughly $500 million in revenue. That is a 240x price-to-sales ratio. The dot-com bubble peaked at 40x.
Two hundred and forty. That is the ratio of aggregate investor valuation to actual revenue across the 10 most valuable pure-play humanoid robot companies, based on confirmed funding rounds, public market caps, and the first credible industry revenue estimate from Counterpoint Research. In 2025, global humanoid robot shipments hit roughly 13,000 units according to Omdia's January 2026 report (subsequent revisions put the total between 13,000 and 14,500). Counterpoint pegs industry revenue at just over $500 million. Investors have decided that same industry is worth north of $120 billion.
For context, Amazon at its 1999 peak traded at roughly 40x sales. OpenAI's latest round valued it at 39x its $4 billion annualized revenue. What the humanoid robot sector is doing right now makes both look conservative.
Who Is Worth What
Start with the numbers. Figure AI confirmed a $39 billion post-money valuation in September 2025 after raising over $1 billion in its Series C, backed by Microsoft, OpenAI, Nvidia, Intel, and Jeff Bezos. The company has two robots operating at BMW's Spartanburg plant. Neura Robotics in Germany targeted $11.6 billion in a Tether-led round. 1X Technologies, the Norwegian-American company backed by OpenAI's startup fund, sought $10 billion or more in September 2025, a 12-fold leap from January 2024.
On the public side, UBTECH Robotics carries a $7.93 billion market cap on the Hong Kong Stock Exchange. Rainbow Robotics, Samsung's strategic partner, trades at $5.81 billion on South Korea's KOSDAQ.
Chinese IPO filings paint a similar picture. Unitree targets a $7 billion valuation on Shanghai's STAR market. AgiBot (Zhiyuan Robotics), the global revenue leader at roughly $140 million in 2025, plans a Hong Kong listing at up to $6.4 billion. Apptronik closed a $403 million Series A at an estimated $5-5.5 billion valuation. Agility Robotics raised $641 million at a $2.12 billion valuation.
Add those 10 companies: $39 + $25 (Boston Dynamics, midpoint of Korean analyst range) + $11.6 + $10 + $7.93 + $7 + $6.4 + $5.81 + $5.25 + $2.12 = $120.1 billion. This excludes Tesla's Optimus (not separately valued), Sanctuary AI, Physical Intelligence, Skild AI, and dozens of smaller companies.
Revenue Reality
Against that $120 billion, the entire industry generated roughly $500 million in revenue last year. Counterpoint's breakdown reveals where the money actually comes from:
| Use Case | % of 2025 Revenue |
|---|---|
| Entertainment & performances | 26% |
| Data production (AI training) | 22% |
| Intelligent manufacturing | 17% |
| Services & guidance | 15% |
| Warehousing & logistics | 6% |
| Other | 14% |
Read that carefully. A quarter of all humanoid robot revenue in 2025 came from entertainment and performances. Another 22% was data production, meaning companies buying robots primarily to collect training data for their own AI models. Only 17% came from manufacturing, the use case most investors cite when explaining their thesis. Warehouse and logistics, the segment Agility and Figure target, contributed 6% of revenue.
AgiBot alone accounted for $140 million, roughly 28% of global revenue. Unitree and UBTECH combined for another 28%. All six of the top shippers by unit volume were Chinese. Figure AI, the most valuable company in the space, does not disclose revenue.
A Calculation Nobody Has Run
I compiled aggregate valuations and revenue for every company where both figures are available. Here is how the humanoid sector's price-to-sales (P/S) ratio compares to other technology sectors at their most optimistic:
| Sector / Era | Aggregate P/S at Peak | Revenue Base |
|---|---|---|
| Humanoid robots (2025) | ~240x | ~$500M actual |
| Dot-com (1999-2000) | ~40x (Amazon), up to 200x (outliers) | Billions actual |
| AV SPACs (2020-2021) | 50-100x | Mostly projected |
| Clean energy SPACs (2020) | 100-500x | Near-zero actual |
| AI/LLM (2024) | ~39x (OpenAI at $157B/$4B) | $4B+ actual |
At 240x aggregate revenue, the humanoid sector exceeds the autonomous vehicle SPAC bubble. It sits closer to the clean energy SPAC era, where QuantumScape traded at 500x revenue on functionally zero sales. Humanoid companies at least have $500 million in actual revenue. That makes the ratio more meaningful, and more alarming.
AgiBot, the revenue leader, is pricing its IPO at roughly 46x its 2025 revenue. That is aggressive but within the range of high-growth SaaS companies (CrowdStrike traded at 40-50x revenue in 2020). Figure AI at $39 billion on undisclosed revenue is in a different category entirely. If Figure generated $50 million in 2025 (a generous estimate given its limited deployments), its P/S is 780x. If it generated $10 million, the multiple is 3,900x.
Capital Goes In, Robots Don't Come Out
Global robotics investment surpassed $10 billion in 2025, with capital concentrating at the late-stage humanoid layer. That money finances hardware R&D, factory buildouts, and AI training infrastructure. It does not, for the most part, finance revenue-generating operations at commercial scale. Agility's RoboFab in Salem, Oregon, can produce up to 10,000 units per year. It currently ships hundreds. Figure plans to produce 100,000 robots "over the forthcoming years." It currently has two working at BMW.
Chinese Companies Actually Ship; Western Companies Get Valued
Omdia's data reveals a striking geographic split. All six of the top humanoid robot shippers in 2025 were Chinese. AgiBot shipped 5,168 units, capturing 39% of global market share. Figure AI, Agility Robotics, and Tesla each held single-digit market share by unit volume. Counterpoint confirms AgiBot, Unitree, and UBTECH collectively generated 56% of global revenue.
Yet the valuation map is nearly inverted. Figure AI ($39B) alone is worth more than AgiBot ($6.4B), Unitree ($7B), and UBTECH ($7.93B) combined. Western companies carry roughly 70% of aggregate industry valuation on perhaps 15% of unit shipments. Chinese companies generate the revenue. American companies attract the capital.
Limitations
Most humanoid companies are private, so "valuations" reflect preferred share prices with liquidation preferences, not public-market clearing prices. Preferred shares typically overstate common-equity value by 20-50%. Revenue figures for Chinese companies are self-reported or analyst-estimated; independent audits are unavailable. Omdia's 13,000-unit figure includes robots used for data collection and entertainment, not productive labor. Tesla's Optimus is excluded because it has no separate valuation, but Musk's claim that it alone justifies Tesla's $800B+ market cap would push the aggregate far higher.
Why Investors Might Be Right
Goldman Sachs Research has modeled scenarios ranging from a $6 billion humanoid market in 2035 (base case) to $154 billion by 2035 if product design, affordability, and public acceptance hurdles clear. Omdia projects 2.6 million annual shipments by 2035, a 200x increase from 2025. If that trajectory holds, this year's revenue is the seed round of a multi-trillion dollar labor market.
Early-stage S-curve pricing has been vindicated before. Waymo generated near-zero revenue until recently and is now valued at $45 billion. Amazon traded at 40x revenue in 1999 and returned 100x. Investors paying 240x today are betting this is 1999 Amazon, not 1999 Webvan.
The Bottom Line
In 2025, the humanoid robot industry generated $500 million in revenue, shipped roughly 13,000 units, and earned a collective valuation exceeding $120 billion. That 240x price-to-sales multiple is not a typo. It is a bet that the industry will grow 200-fold in a decade, that robots now performing stage shows and collecting training data will be manufacturing cars and staffing warehouses at scale, and that companies currently operating a few dozen robots will be operating millions. Every dollar invested today buys five cents of annual revenue. For this to work, humanoid robots need to become the most successful industrial product category since the semiconductor. History says most 240x bets lose. But the ones that win rewrite the global labor market.