Signals Inbox·August 21, 2026·Humanoid Robotics

Is Unitree really worth $50B today?

No: Unitree has built one of the strongest humanoid manufacturing businesses in the world, but a $50B valuation still prices in several years of revenue growth, industrial adoption and software progress that have not arrived yet.

Q1What exactly happened when Unitree hit $50B?

Unitree did reach a $50.7 billion valuation on its first day of trading, but that number lasted barely a session: the company is now worth roughly $40 billion after another sharp pullback.

The sequence is almost as interesting as the valuation itself. Unitree’s last major private financing valued the company at about RMB12 billion, or $1.7 billion, in 2025. Investors in that round included Geely, Ant Group and HongShan. The IPO was then priced at RMB150.80 per share, implying a market value of about RMB61 billion, close to $9 billion.

Then trading began. Unitree opened at RMB1,100, briefly pushing its valuation above RMB440 billion, before closing its first session at RMB845 and RMB341.7 billion, or $50.7 billion. The stock fell 18.7% the next session and closed the latest session at RMB672.41. Based on the 404.5 million shares outstanding after the IPO, that leaves Unitree around RMB272 billion, roughly $40 billion.

So we have three very different valuations within the same story: $1.7 billion in its private round, about $9 billion at the IPO price, and more than $50 billion at the first close. Even after the latest correction, Unitree is still worth more than 20 times its private valuation from little more than a year ago.

The IPO mechanics also tell us how overheated the debut became. Investors submitted orders for thousands of times the shares available, nearly 9.8 million retail accounts participated in the online subscription, and 85% of Unitree’s freely traded shares changed hands on the first day.

There is an even cleaner benchmark. Reuters reported shortly before the IPO that lead underwriter CITIC Securities valued Unitree at RMB50.6 billion to RMB55.9 billion over the following six to twelve months. The first-day close was more than six times the top of that range. Even after the latest fall, Unitree trades at almost five times it.

That gap is where the real valuation debate starts.

Q2Does Unitree actually have enough revenue for a $50B valuation?

No. Unitree’s revenue is real and growing fast, but $50 billion meant paying roughly 200 times its latest audited annual sales.

Unitree’s IPO prospectus gives us unusually solid numbers. Revenue increased from RMB159 million in 2023 to RMB393 million in 2024 and RMB1.699 billion in 2025, around $250 million. That means the business grew 147% in 2024 and another 333% in 2025.

At the $50.7 billion first-day close, investors were therefore paying about 201 times 2025 revenue. Even at the current valuation near $40 billion, the multiple is still around 160 times.

Those are extraordinary numbers even for software companies with recurring subscriptions and very high margins. Unitree sells physical robots, has manufacturing costs, depends on suppliers and has to keep improving hardware every generation.

The IPO price itself gives us a useful contrast. At RMB150.80 per share, Unitree was valued at about RMB61 billion, or roughly 36 times 2025 revenue. That was already expensive. Public trading then multiplied the company’s valuation more than fourfold without the underlying revenue changing.

CITIC’s pre-IPO work offers another reality check. Its RMB50.6–55.9 billion valuation corresponded to roughly 20 times expected 2026 sales. Investors at the first-day close were effectively paying more than six times the equity value that Unitree’s own IPO sponsor had just argued was reasonable.

A large premium for Unitree’s growth is defensible. A 160–200x sales multiple requires something much stronger.

Q3Is Unitree still growing fast enough today?

Unitree is still growing quickly, but the latest numbers show a major slowdown from the explosive pace that created the valuation story.

Revenue jumped 333% in 2025. In the first quarter of this year, revenue reached roughly RMB420 million, up about 69% from a year earlier. Unitree then guided to first-half revenue of RMB1.052–1.128 billion, implying growth of 35.6–45.4%.

Going from 333% annual growth to roughly 40% growth does not make Unitree a slow company. Plenty of excellent businesses would love to grow 40%. The problem comes from matching that growth rate with a valuation above 150 times trailing revenue.

The robot volumes are growing faster than revenue. Unitree shipped more than 5,500 humanoids in 2025, after shipping only 537 in 2024. Its quadruped volumes also jumped from 7,136 units to more than 23,000.

The reason revenue grows more slowly is pricing. Unitree’s humanoid average selling price fell from roughly RMB260,000 in 2024 to RMB166,000 in 2025. So Unitree can sell many more robots while getting less revenue from each one.

Two things are happening at once: Unitree is currently scaling robot production extremely fast, while revenue growth has already moved into a much lower range than last year. A $50 billion valuation needs that second curve to stay exceptional too.

Q4Has Unitree really become a humanoid robot company?

Yes. Humanoids went from a tiny experiment inside Unitree to its largest revenue business in only two years.

In 2023, humanoid robots contributed just 1.9% of Unitree’s main-business revenue. That jumped to 27.7% in 2024 and 51.8% in 2025.

The absolute change is even clearer. Humanoid revenue went from about RMB3 million in 2023 to RMB107 million in 2024 and RMB868 million in 2025. Recognized unit sales rose from five robots to 412 and then 5,215.

Unitree still has a large quadruped business. It sold more than 23,000 quadrupeds in 2025, generating roughly RMB698 million. Yet humanoids produced more revenue despite Unitree selling more than four robot dogs for every humanoid.

That transformation happened remarkably fast. The H1 humanoid was only launched in 2023. Two years later, humanoids generated more than half of company revenue.

For the valuation, this is genuinely important evidence. Investors are not taking an old robot-dog manufacturer and attaching a humanoid story to it. The revenue mix has already changed.

Q5Who is actually buying Unitree’s humanoid robots right now?

Unitree has thousands of paying humanoid customers, but current demand is still heavily concentrated in research, education and early-stage robotics development rather than large-scale industrial work.

The Financial Times recently reported that research, education and government-supported use remain major sources of demand across China’s humanoid industry. Unitree’s own disclosures similarly show universities, research institutes, technology companies and developers as important customers.

That distinction has become more relevant lately. Chinese local governments are funding robot training centers that buy humanoids, collect teleoperated task data and sometimes sell that data back into the robotics ecosystem. The Financial Times found enough of this activity to raise a legitimate question about how much current Chinese demand represents normal commercial purchasing by companies trying to cut labor costs.

Unitree’s founder Wang Xingxing has been unusually open about the commercial gap. Speaking recently at the World Robot Conference, he explained why Unitree has not deployed robots across factories and homes at much larger scale: today’s machines still lack the efficiency and capability to compete with humans across many tasks. He estimated that reaching the point where humanoids can handle most unfamiliar instructions could take anywhere from two to ten years.

That comment deserves more weight than another dancing or boxing demonstration. Unitree has already solved enough problems to sell thousands of robots. The next step is harder: customers need to operate them every day because the robots save money or perform work humans cannot.

We have much less evidence of that today.

Q6How expensive is Unitree compared with public robotics companies?

Unitree currently trades at a valuation that public robotics companies do not come close to matching on revenue.

The comparison needs some care. FANUC sells mature industrial robots. Intuitive Surgical dominates robotic surgery. Symbotic automates warehouses. UBTECH is the closest listed humanoid comparison. None is growing exactly like Unitree, so Unitree deserves a premium.

The size of the premium is still striking. UBTECH generated about RMB2.0 billion of 2025 revenue, slightly more than Unitree, and is worth roughly $5.5 billion. Its humanoid revenue also accelerated dramatically, reaching about RMB821 million in 2025. Yet Unitree is currently valued at roughly seven times more.

Symbotic generated about $2.65 billion of trailing revenue, more than ten times Unitree’s 2025 revenue, while its market capitalization recently sat near $25 billion. Intuitive Surgical produces more than $11 billion of annual revenue, grows around 20% and trades near 13 times sales.

Unitree can reasonably trade above these companies on sales because its growth opportunity is much earlier. Going from 13–20x to around 160x requires us to believe its eventual economics will be radically better.

Unitree versus public robotics companies

Q7Is Unitree really worth more than Figure AI?

We struggle to justify valuing Unitree meaningfully above Figure AI today because the two companies have proved different parts of the humanoid thesis.

Figure raised more than $1 billion at a $39 billion post-money valuation in its latest major round. Unitree is now trading in roughly the same range after its pullback, while its $50.7 billion debut temporarily put it well above Figure.

Unitree has the stronger manufacturing evidence. It already sells thousands of humanoids, generated RMB1.7 billion of audited revenue last year and can manufacture robots at prices that make widespread experimentation possible.

Figure has shown more convincing industrial deployment evidence. Its Figure 02 robots worked at BMW’s Spartanburg factory for roughly ten months, ran ten-hour weekday shifts, logged more than 1,250 operating hours, loaded over 90,000 parts and contributed to production of more than 30,000 BMW X3 vehicles. Figure has since moved a new generation of robots into BMW.

That BMW deployment gives us something Unitree still lacks at comparable scale: a documented example of humanoids repeatedly participating in a live high-value manufacturing process.

Figure also puts much more of its strategic emphasis on its Helix AI models. Unitree is investing heavily in embodied AI too, including roughly RMB2 billion of planned IPO proceeds for intelligent robot model R&D, but its visible advantage today remains manufacturing.

At around $40 billion each, we can understand the argument for similar valuations. Figure offers stronger evidence around autonomy and industrial deployment; Unitree offers stronger revenue, volume and manufacturing evidence.

At $50 billion, Unitree was being priced as though it had already won the comparison. The evidence is not there yet.

Q8Why is Unitree worth so much more than Apptronik and Agility Robotics?

Unitree deserves to be worth more than Apptronik and Agility today, but the current gap still looks unusually large.

Apptronik expanded its Series A to $935 million this year. TechCrunch reported a post-money valuation around $5.3 billion. Its Apollo humanoid program has relationships with Mercedes-Benz, Jabil, GXO, John Deere and Google DeepMind.

Agility Robotics has agreed to go public through a transaction valuing it at $2.5 billion before new money. Agility says its Digit robots have accumulated more than 65,000 operating hours across nine customer facilities, and the company has secured more than $300 million of multi-year Digit v5 orders, subject to deployment milestones.

Those figures make the comparison uncomfortable for Unitree shareholders. At roughly $40 billion today, Unitree is valued at around eight Apptroniks and sixteen Agility Robotics.

Unitree has far more current unit volume and real reported revenue, so a meaningful premium makes sense. Agility, however, can point to more than $300 million of contracted future orders at a $2.5 billion valuation. Unitree generated roughly $250 million of total revenue last year and trades near $40 billion.

The market is assigning a very large value to Unitree’s manufacturing scale, Chinese supply chain, brand and potential leadership. The logic is visible. The magnitude still looks aggressive.

Humanoid company valuation comparison

Q9Is Unitree actually leading the humanoid robot market right now?

Unitree is clearly one of the world’s largest humanoid manufacturers, although calling it the undisputed global number one today goes too far.

Its IPO prospectus says Unitree shipped more than 5,500 humanoids in 2025 and ranked first globally. That alone puts it far ahead of most Western competitors on physical production volume.

More recent third-party estimates show a tighter race. Smart Analytics Global estimated that AgiBot captured roughly 44% of global humanoid shipments in the first half of this year, with Unitree around 31%. Business Insider separately cited the same 31% figure for Unitree.

The precise rankings should be treated carefully because definitions differ. Some datasets count shipments, others sales, and companies classify wheeled humanoids or research platforms differently.

The broader pattern is much harder to dispute: Unitree and AgiBot have already reached shipment volumes that most Western humanoid companies have not.

For Unitree’s valuation, being top two rather than definitively number one changes little. The company has already proved it can manufacture humanoids at meaningful volume. The unanswered question is who will dominate the much larger market for robots doing valuable work every day.

Q10Is the humanoid robot market getting big fast enough?

The humanoid robot market is growing extremely fast from a tiny base, while commercial demand is still years behind the most aggressive long-term forecasts.

Recent industry estimates put first-half humanoid shipments around 19,000 units, nearly four times the level from a year earlier. China accounts for the vast majority of those shipments.

Money is arriving even faster. Chinese embodied-AI companies attracted roughly RMB93.5 billion, close to $14 billion, of financing in the first half of this year according to ITJuzi data reported by Yicai. That was about five times the previous year, across 322 financing deals.

Put those together: robot volumes are exploding, funding is exploding and new companies are appearing everywhere. China alone has produced hundreds of embodied-AI startups in a short period. Those conditions can create enormous companies, and they can also produce a lot of capital chasing the same opportunity before commercial demand catches up.

The long-term forecasts are huge. UBS has estimated more than two million humanoids in operation by 2035 and more than 300 million by 2050. Its projected market reaches roughly $40 billion around 2035 before moving toward $1.4–1.7 trillion by 2050. Morgan Stanley has modeled an even larger humanoid economy above $5 trillion by 2050.

The awkward comparison is that Unitree briefly reached a $50 billion equity valuation before the entire humanoid market, under UBS’s forecast, reaches roughly $40 billion of annual size around 2035.

That does not make the valuation mathematically impossible. Equity value reflects many years of future profits. It does show how far into the future today’s price is looking.

Q11What does Unitree do better than competitors?

Unitree’s clearest advantage today is its ability to build capable robots cheaply, in volume and with unusually deep control over the hardware stack.

The company develops major pieces internally, including motors, reducers, controllers, perception systems, joint modules, dexterous hands and motion-control technology. Industry analysis around the IPO estimated that roughly 90% of its supply chain is domestic, while core components are internally developed.

The best proof is in Unitree’s pricing and margins. Its average humanoid selling price fell from about RMB593,000 in 2023 to RMB260,000 in 2024 and RMB166,000 in 2025. Over the same period, company-wide core-business gross margin increased from roughly 44% to 60%.

Companies usually find it difficult to cut prices that aggressively while improving their economics. Unitree managed both while multiplying production.

That gives the company a useful flywheel. Lower prices bring more universities, developers and companies onto Unitree hardware. More deployments create product feedback and potentially more training data. Higher volumes then give Unitree more purchasing and manufacturing leverage.

The weaker part of the moat is software. Figure is pushing Helix, Google DeepMind is working with Apptronik, and many Chinese competitors are building their own embodied foundation models. Unitree is investing heavily here, but we have not yet seen evidence that its autonomy software enjoys the same lead as its hardware.

Today, Unitree’s manufacturing advantage deserves a premium. Assuming it already owns the intelligence layer is much harder.

Q12Is Unitree actually making good money on its robots?

Yes. Unitree’s hardware economics are surprisingly good today, and this is one of the strongest arguments for giving the company a premium valuation.

Unitree reported RMB278 million of net profit on RMB1.699 billion of revenue in 2025. That works out to a reported net margin around 16%. For a young robotics manufacturer scaling this quickly, being profitable at all is unusual.

Gross margins are even stronger. Core-business gross margin reached about 60%. Unitree’s humanoid gross margin was around 63% in 2025.

The margin trend needs a little more nuance. Humanoid gross margin has fallen from extremely high early levels as prices came down and competition increased. Yet Unitree has compensated through scale and manufacturing efficiency well enough to keep overall company gross margin moving upward.

The first quarter also showed why we should avoid assuming every year will look like 2025. Revenue increased about 69%, while reported profit fell sharply as Unitree spent more on R&D, engineering talent, product development and marketing.

That spending does not worry us much by itself. A company at this stage should be investing. More importantly, Unitree has already shown it can manufacture complex robots without burning huge amounts of cash on every unit.

Still, keep the scale of the valuation in view. RMB278 million of reported annual profit against a current market value near RMB272 billion means investors are paying close to 1,000 times last year’s earnings. Current profits validate the business model; they do very little to validate the stock price.

Q13Could U.S. restrictions cut into Unitree’s growth?

Yes, and the exposure is meaningful because 43.7% of Unitree’s main-business revenue came from overseas customers last year.

The United States has recently tightened restrictions around foreign advanced robotics. The FCC added foreign-produced advanced robotic devices to its Covered List, making new equipment authorizations much harder without specific approval. The Pentagon has also added Unitree to its list of Chinese companies it says assist China’s military.

That does not mean 44% of revenue suddenly disappears. Unitree’s overseas revenue covers many countries, and the United States represents only one part of it.

The bigger issue is strategic. Part of the $50 billion case assumes that Unitree can become one of the default humanoid platforms globally. Restrictions in the United States can block access to some of the world’s most valuable enterprise, government and critical-infrastructure customers.

They also create an opening for Figure, Apptronik, Agility and other Western companies. A multinational manufacturer may prefer a robot platform it can deploy across factories in several countries without worrying about future restrictions.

Unitree can still build a huge business across China, Asia, the Middle East, Europe and other markets. The global opportunity available to it is simply less frictionless than the headline humanoid TAM suggests.

Q14How much revenue would Unitree need to make $50B look normal?

Unitree probably needs at least $1.7–2.5 billion of annual revenue before a $50 billion valuation starts to look defensible on recognizable high-growth multiples.

At 30 times revenue, a $50 billion company needs about $1.67 billion of sales. At 20 times, it needs $2.5 billion. At 10 times, it needs $5 billion.

Those multiples are already generous. A 20–30x sales valuation would put Unitree above almost every major public robotics company today.

Starting from approximately $250 million of 2025 revenue, Unitree needs to become roughly seven times larger just to reach the revenue required at 30x sales. The 20x case requires ten times more revenue.

CITIC’s own valuation work is useful here. Its 20x expected-sales framework produced only around $7.4–8.3 billion of equity value. To reach $50 billion using a similar multiple, Unitree would need several billion dollars of revenue rather than several hundred million.

If Unitree maintained 40% annual revenue growth from the 2025 base, getting to $2.5 billion would take close to seven years. At 70% growth, it would take a little over four years. Sustaining 70% for four years in hardware would itself be an exceptional outcome.

Revenue required to support a $50B valuation

Q15What would have to go right for Unitree to grow into $50B?

Unitree can eventually justify $50 billion, but we would need to see industrial demand catch up with its manufacturing success.

Revenue growth is the first test. Roughly 40% growth is excellent in isolation; at today’s valuation, we would want stronger growth or clear evidence that another acceleration is coming.

The quality of customers then needs to change. Research institutes and developers helped Unitree scale quickly. The next several thousand robots should increasingly appear inside factories, warehouses and commercial operations where someone can show the cost saved per task, the uptime achieved and the payback period.

Software could change the economics completely. If Unitree eventually sells autonomy, fleet management, updates, applications, maintenance and other recurring services around its low-cost hardware, revenue per deployed robot could rise even as hardware prices keep falling.

Its current 60% core-business gross margin gives Unitree room to pursue that model. The company also raised about RMB6.1 billion through the IPO and plans to spend heavily on robot models, robot bodies, new products and manufacturing capacity.

Then there is scale. Unitree currently produces thousands of humanoids. A $50 billion outcome becomes much easier to defend if we start talking about tens of thousands, then hundreds of thousands, while margins remain healthy.

There is a real path to $50 billion. Unitree has to become a large industrial and AI platform built on top of its manufacturing lead, rather than simply sell far more inexpensive robots.

Q16What could crush Unitree’s valuation even if the company keeps growing?

The easiest way for Unitree’s valuation to break is surprisingly mundane: revenue keeps growing 30–40% while investors stop paying triple-digit sales multiples.

Unitree would still be a successful company in that scenario. Suppose revenue doubles over the next few years but the valuation falls toward 30 times sales. The business could become substantially larger while the stock remains flat or drops.

Price competition creates another problem. Unitree has handled falling prices remarkably well so far, but China now has hundreds of robotics startups competing for capital, engineers and customers. UBTECH, AgiBot, Galbot, Deep Robotics and many smaller companies are all trying to scale.

Commercial adoption could also take longer than the market expects. As Unitree’s own founder recently acknowledged, today’s robots still have large efficiency and capability gaps against human workers. If that problem takes closer to ten years than two years to solve, the valuation is discounting the opportunity far too early.

Software is another pressure point. Unitree could dominate robot manufacturing while a different company captures the most valuable autonomy layer. Hardware suppliers rarely earn the same economics as the platform controlling the software ecosystem.

Geopolitical restrictions add one more source of pressure, especially with such a large overseas business.

None of these outcomes requires Unitree to become a bad company. That is exactly why we find the valuation risky. Great companies can still be terrible purchases when the price already assumes several future victories.

Q17So is Unitree really worth $50B today?

No. We think Unitree is one of the most impressive humanoid robotics companies operating today, but the evidence still puts $50 billion ahead of the business by several years.

The IPO makes that easier to say with confidence. Unitree has given us audited numbers: RMB1.7 billion of annual revenue, more than 5,500 humanoid shipments, roughly 60% core-business gross margin and positive net income. Humanoids already generate more than half of revenue. The company has also proved it can bring prices down dramatically while scaling manufacturing.

Those are excellent fundamentals.

The valuation asks for much more. Unitree briefly traded around 201 times 2025 revenue. Even after falling from $50.7 billion to roughly $40 billion, the multiple remains around 160 times. UBTECH trades around 20 times sales. Symbotic is near 9 times. Intuitive Surgical is around 13 times. CITIC Securities valued Unitree around $7–8 billion shortly before the IPO using roughly 20 times expected 2026 revenue.

And current commercial reality is still early. Research and education remain important buyers. China’s government-backed robotics ecosystem is helping stimulate demand. Large autonomous industrial deployments remain limited. Unitree’s founder himself says current robots are still less efficient than humans for many jobs.

Our view is therefore fairly simple. Around $50 billion, Unitree was clearly stretched. Around $40 billion today, it is still stretched, although the case is less extreme after the pullback.

We would become much more comfortable near $50 billion once Unitree is producing roughly $1.7–2.5 billion of annual revenue, keeping strong margins, deploying robots at real industrial scale and showing that its embodied-AI software is becoming an advantage rather than simply another expensive R&D program.

Unitree may get there. Its manufacturing progress gives us a credible reason to take that possibility seriously. Investors currently have to pay for much of that future before it has arrived.