What are the most valuable Physical AI startups?

Top startups in the Physical AI Market ranked by valuation

Here is an updated table that ranks the top startups in the Physical AI Market based on their latest reported or estimated valuations.

If you need to dig deeper and get more detailed data, please check our report covering the Physical AI Market.

Is Waymo really worth $126 billion?

Yes. Waymo’s $126 billion valuation is one of the strongest valuation figures anywhere in the Physical AI market today.

Waymo announced a $16 billion financing at a $126 billion post-money valuation, giving us a completed transaction rather than an inferred multiple or an active fundraising target. In a market where many of the largest numbers still come from negotiations, secondary transactions or estimates, that is a much cleaner anchor.

Waymo also has operating scale that most Physical AI startups do not yet have. The company has moved beyond testing autonomous vehicles into commercial robotaxi operations across multiple U.S. markets, while Alphabet remains its majority shareholder.

A $126 billion valuation is enormous even by technology-startup standards, but investors are valuing Waymo against a transportation market measured in trillions of dollars rather than against today's robotics revenue alone. If autonomous driving captures a meaningful part of ride-hailing, private transportation, logistics and vehicle software, Waymo is already in one of the strongest technological and operational positions.

Why is Waymo worth so much more than humanoid robot startups?

Waymo is worth much more than humanoid robot startups because autonomous driving has already crossed several commercial hurdles that humanoid robotics is still trying to clear.

Waymo operates machines that perform a valuable human task today: driving passengers. The company has accumulated years of real-world autonomous miles, regulatory experience, fleet operations and consumer usage. Investors have far more evidence about technical performance, demand and deployment economics.

Figure AI, Apptronik, 1X and other humanoid companies face a broader opportunity but a much earlier commercial market. A general-purpose humanoid could eventually work across warehouses, factories, retail stores and homes. The problem is that reliably doing hundreds of different physical tasks for thousands of hours remains much harder than producing a convincing demonstration.

The valuation gap captures that difference. Figure sits around $39 billion to $44 billion in our dataset, while Apptronik is near $5.5 billion. Waymo reaches $126 billion because autonomous driving already looks much more like an operating business.

Is Anduril really worth around $100 billion now?

A valuation close to $100 billion for Anduril should currently be treated as an aggressive active-raise estimate, while the strongest completed financing evidence supports a lower figure.

Anduril’s rise is still exceptional. A completed financing earlier this year raised $5 billion and valued the defense-technology company at roughly $61 billion, almost doubling its previous valuation. The Financial Times also reported that Anduril generated about $2.2 billion of revenue in 2025.

Our dataset uses a $95 billion to $105 billion active-raise valuation, reflecting a newer implied market level rather than the last firmly completed round. That may eventually prove accurate, especially if a new financing closes there, but we would not give it the same confidence as Waymo's $126 billion or Saronic's $9.25 billion.

The direction is easier to defend than the exact number. Anduril has moved from venture-backed defense startup to a serious challenger to established contractors. Its Lattice software, autonomous aircraft, counter-drone systems and manufacturing expansion give it exposure to several parts of the modernization of Western military procurement.

Defense tech is also being repriced more broadly. Venture investment in the sector reached about $12.3 billion during the first half of 2026 according to the Financial Times, already exceeding the amount invested during all of 2025.

Is Figure AI really worth nearly $40 billion?

Yes. Figure AI has a directly disclosed $39 billion post-money valuation, although moving much above that level requires considerably more interpretation.

Figure announced more than $1 billion of committed Series C capital at a $39 billion post-money valuation in 2025. That remains a strong transactional anchor for the humanoid company.

The unusual part is how early Figure reached that number. The company is valued above mature robotics businesses that have spent years shipping machines, and its valuation is roughly seven times Apptronik's most recently disclosed level.

Investors are effectively paying for Figure to become both a major robot manufacturer and a major AI company. Figure is building its Helix intelligence system, collecting large volumes of humanoid training data and expanding BotQ manufacturing. More recently, it signed a commercial deployment agreement with Catalyst Brands and announced a compute partnership with Nscale that could involve up to 100,000 Nvidia Vera Rubin GPUs.

At $39 billion, becoming a successful niche robot producer would not be enough. Figure needs to become a very large platform.

Is Unitree now one of the world’s most valuable robotics companies?

Yes. Unitree has become one of the largest publicly valued pure-play robotics companies in the world, although its market value has already shown how quickly Physical AI enthusiasm can reprice a stock.

Unitree's Shanghai IPO became one of the most extraordinary robotics listings yet. The company raised about $904 million, and Bloomberg reported that its shares surged roughly 460% on their first day, producing a closing market capitalization near 342 billion yuan.

That was around $50 billion at the time, substantially above the $27.2 billion current-market snapshot used in our dataset.

The difference is useful rather than contradictory. Unitree is public, so its valuation can move dramatically from one week to another. A funding-round valuation stays fixed until another transaction occurs; a market capitalization does not.

Unitree also gives investors something many humanoid startups still lack: material unit sales. Financial Times reporting indicated that the company sold about 28,000 robot dogs and more than 5,000 humanoid robots in the prior year.

The same reporting showed why investors should remain careful. Unitree traded at more than 150 times expected 2026 sales around its early public-market peak. That is an extraordinary multiple for a hardware-heavy business.

How many Physical AI unicorns are there now?

Physical AI has already produced roughly 60 companies around or above unicorn territory in our dataset, which shows how far the investment boom has spread beyond a handful of famous humanoid startups.

Fifty-eight companies in the ranking have valuations that are clearly at or above $1 billion. Another group including Reliable Robotics, Lightwheel, Sereact, Booster Robotics, Oxa and Dexory have ranges that reach or cross $1 billion.

The count is broad-based. Billion-dollar valuations now exist across almost every serious Physical AI category.

Autonomous driving produced Waymo, Wayve, Waabi and Nuro. Humanoids produced Figure, Unitree, Apptronik, 1X, NEURA, XPENG Robotics and several Chinese challengers. Defense autonomy produced Anduril, Shield AI, Saronic, Quantum Systems, TEKEVER and others. Industrial robotics produced Dexterity, Pudu, Mujin, RobCo and Standard Bots. Robot intelligence produced Skild AI, Physical Intelligence, FieldAI, Generalist AI and Rhoda AI.

A $1 billion Physical AI startup can now rank outside the top 50.

Which parts of Physical AI attract the highest valuations?

Autonomous driving still produces the largest single Physical AI valuation, while humanoids, defense autonomy and robot foundation models dominate much of the rest of the upper end.

Waymo alone gives autonomous mobility a huge lead. Wayve, Applied Intuition, Waabi and Nuro add several more multibillion-dollar companies.

Humanoid robotics has more companies crowded near the top. Figure, Unitree, NEURA, XPENG Robotics, 1X and Apptronik all sit above roughly $5 billion in our dataset.

Defense autonomy is similarly strong. Anduril, Shield AI, Saronic, Quantum Systems and TEKEVER together represent well above $100 billion using the estimates in our ranking, even before adding smaller drone, maritime and ground-autonomy companies.

Then there is the newer intelligence layer. Skild AI, Physical Intelligence and FieldAI are being valued as potential general-purpose operating brains for robots rather than as manufacturers of one specific machine.

Traditional specialized robotics still produces valuable companies, but valuations such as Pudu's roughly $1.5 billion or Dexterity's $1.65 billion remain far below the amounts attached to general-purpose platforms.

Are humanoid robot valuations getting ahead of the actual business?

Yes. Humanoid robot valuations are currently moving much faster than commercial revenue and deployment, even though the technology is becoming more credible.

Figure reached a $39 billion financing valuation before humanoids had become a large established commercial category. Apptronik then raised another $520 million at more than $5.5 billion, roughly tripling the valuation attached to its earlier Series A financing. XPENG Robotics recently raised more than $900 million at a post-money valuation above $6.3 billion before its planned large-scale external robot deliveries.

Unitree provides the clearest public-market comparison. Unlike most humanoid startups, it already sells thousands of machines, yet the Financial Times calculated that the company traded above 150 times expected 2026 revenue around its post-IPO surge.

Those prices only work if humanoid robots become a much larger market than they are today.

There are reasons investors believe that can happen. Improvements in vision-language-action models are making robots easier to train. Companies can reuse factories, warehouses and homes designed around human bodies. Manufacturing costs are gradually falling. Large companies including Mercedes-Benz, GXO, BMW, Amazon-linked investors and Google have also become involved in deployments or partnerships.

The hard part is scale. A robot that completes a task during a pilot still has to work reliably, cheaply and safely for thousands of hours before an employer replaces or supplements a human workflow with it.

Are robot foundation-model companies becoming as valuable as robot manufacturers?

Yes. Investors are increasingly valuing the companies building robot intelligence almost like major robot manufacturers, and Skild AI is the strongest example.

Skild AI raised $1.4 billion at a valuation above $14 billion after being valued at $1.5 billion in its 2024 Series A. That is roughly a ninefold valuation increase in about eighteen months.

The commercial evidence has also started to improve. Skild says it crossed $100 million in annual recurring revenue only ten months after its first commercial deployment. It has partnerships involving ABB Robotics, Universal Robots and Nvidia, and it acquired Zebra Technologies' robotics arm, formerly Fetch Robotics.

Physical Intelligence follows the same thesis. The company raised $600 million at a $5.6 billion post-money valuation in late 2025. Our dataset pushes its current estimate higher as the market for robot foundation models has repriced, but that additional appreciation should still be treated as inferred until another transaction confirms it.

Generalist AI raised another $400 million this year, bringing its total financing above $500 million. Its GEN-1 system is designed to transfer intelligence across robots, tasks and end effectors rather than remain attached to one machine.

Rhoda AI offers another variation. Bloomberg reported that it raised $450 million at a $1.7 billion valuation to develop models that learn industrial robot behavior from large amounts of video.

Is defense autonomy becoming the biggest Physical AI market after autonomous driving?

Defense autonomy is currently the strongest cluster of high-valued Physical AI companies after autonomous mobility, and its valuations are being supported by unusually powerful demand.

The category contains several different businesses. Anduril combines software, drones, sensors and manufacturing. Shield AI develops autonomous aircraft and its Hivemind AI pilot. Saronic builds autonomous naval vessels. Quantum Systems develops multi-domain unmanned systems. TEKEVER produces autonomous aerial systems used for defense and intelligence.

These companies are repricing rapidly. Quantum Systems raised $1.2 billion at roughly an $8 billion post-money valuation after being valued above €3 billion only months earlier. TEKEVER raised $580 million at a $6.4 billion valuation after reaching unicorn status the previous year. Saronic raised $1.75 billion at a $9.25 billion valuation.

Here the buyer is obvious: governments.

Wars in Ukraine and the Middle East, U.S. military modernization, European rearmament and demand for cheaper autonomous systems have created budgets that can support large companies.

Company	Main focus	Current valuation anchor	Recent financing evidence
Anduril	Multi-domain autonomous defense	~$61B completed round; higher estimates exist	$5.0B round
Shield AI	AI pilots and autonomous aircraft	$12.7B completed round	$2.0B financing package
Saronic	Autonomous naval vessels	$9.25B	$1.75B Series D
Quantum Systems	Multi-domain autonomous systems	~$8.0B	$1.2B Series D
TEKEVER	Autonomous defense drones	$6.4B	$580M Series D
Skydio	Autonomous drones	$4.4B in our dataset	$854M total funding
STARK	AI-enabled defense drones	$3.8B–$4.2B estimate	~$659M total funding
Allen Control Systems	Autonomous counter-drone systems	$2.2B	$242M total funding

Is China catching the US in Physical AI?

China has already become a serious peer in hardware-heavy Physical AI, particularly humanoids, although the largest American companies still command much higher valuations.

Unitree is the strongest symbol of the change. Its public listing briefly pushed its market capitalization past $50 billion. XPENG Robotics then raised more than $900 million at a valuation above $6.3 billion, calling the round the largest private financing in China's embodied-AI industry.

The depth below those companies is just as important. Our dataset includes Galbot at roughly $3.3 billion to $4.2 billion, X Square Robot around $2.8 billion to $3.1 billion, AI² Robotics in a similar range, LimX Dynamics around $2.2 billion to $2.6 billion, plus RobotEra, EngineAI, Deep Robotics, Leju Robotics and Fourier Intelligence.

China's advantage comes partly from an industrial ecosystem built around motors, actuators, batteries, electronics and high-volume manufacturing. That makes it easier to reduce robot hardware costs and iterate physical products quickly.

The United States remains stronger in the highest-value AI and autonomy companies. Waymo, Anduril, Figure, Skild AI, Physical Intelligence and several defense startups create a very large lead at the top.

China has not overtaken the U.S. Physical AI ecosystem by valuation. It has, however, built a much deeper hardware cohort than someone looking only at Figure and Tesla might assume.

Which Physical AI startups have raised the most money, and does that make them more valuable?

Waymo has absorbed the most capital in our Physical AI dataset by a very wide margin, but total funding does not reliably predict valuation.

Waymo's cumulative funding reaches roughly $27 billion in our data after its latest $16 billion round. Anduril follows above $11 billion. Motional has received about $4 billion, Shield AI around $3.6 billion in our dataset, Wayve roughly $2.6 billion, Saronic around $2.6 billion and Skild AI above $2 billion.

That funding intensity reflects the economics of Physical AI. Training AI models is expensive, but building factories, vehicles, sensors, actuators, batteries, testing infrastructure and fleets pushes capital requirements even higher.

The amount raised still says little on its own about what investors think the company is worth. Motional has received roughly $4 billion while our current valuation estimate sits around $2.8 billion to $3.6 billion. Nuro has raised more than $2 billion but is valued around $3 billion to $4.2 billion.

Figure shows the opposite pattern. Its cumulative funding remains much lower than Waymo's, yet investors valued it at $39 billion because they are pricing a potential leadership position in general-purpose humanoids.

The more useful question is what the capital produced: a technological moat, commercial deployment, scalable manufacturing or a credible route into a very large market.

Company	Approx. total funding in dataset	Current valuation
Waymo	$27.1B	$126.0B
Anduril	$11.35B	$95.0B–$105.0B estimate
Motional	~$4.0B	$2.8B–$3.6B
Shield AI	$3.57B	$19.0B–$22.0B estimate
Wayve	$2.60B	$8.5B–$8.6B
Saronic	$2.58B	$9.3B
Nuro	$2.34B	$3.0B–$4.2B
Skild AI	$2.22B	$14.0B–$16.0B
Physical Intelligence	~$2.12B	$10.5B–$11.5B estimate
Quantum Systems	$1.94B	$7.7B–$8.0B

Which Physical AI companies have increased their valuations the fastest?

Some Physical AI valuations have multiplied several times in little more than a year, showing how quickly investors have repriced robotics and autonomous systems.

Skild AI is one of the most dramatic examples. Its 2024 Series A valued the company at $1.5 billion. Its Series C this year valued it above $14 billion, a roughly ninefold jump.

Shield AI moved from $5.3 billion in early 2025 to $12.7 billion in its completed 2026 financing package. Quantum Systems moved from above €3 billion late last year to roughly $8 billion this year.

Apptronik's latest financing valued it above $5.5 billion, about three times the valuation attached to its earlier Series A. TEKEVER went from just over £1 billion when it was confirmed as a unicorn to $6.4 billion in its latest round.

These increases are too large to be explained by ordinary year-on-year business growth alone. Investors have also changed the multiples they are willing to assign to Physical AI.

Company	Earlier valuation anchor	Newer valuation anchor	Approximate change
Skild AI	$1.5B	>$14B	>9×
TEKEVER	>£1B	$6.4B	roughly 4–5×
Apptronik	roughly $1.8B	>$5.5B	~3×
Quantum Systems	>€3B	~$8B	>2×
Shield AI	$5.3B	$12.7B	~2.4×
Figure AI	$2.6B in 2024	$39B in 2025	~15×
Unitree	~$9B IPO valuation	>$50B at first-day close	>5× intraday/public repricing

Are investors valuing humanoid hype more than actual robot sales?

Right now, investors often value future general-purpose capability more highly than proven robot sales, and Pudu Robotics makes that contrast unusually clear.

Pudu says it has shipped more than 120,000 robots globally. The company reported more than 100% revenue growth in 2025, operates across more than 80 countries and has customers including large retailers. It raised nearly $150 million this year at a valuation above $1.5 billion.

Figure is worth about 25 times as much according to its latest disclosed financing valuation.

Figure's premium is not about current sales. Investors are underwriting a very different outcome.

Pudu's existing business is largely built around commercial cleaning, service delivery and industrial delivery. These are real markets, but investors can already see the shape of the business.

Figure's possible end state is much harder to bound. If one general-purpose humanoid platform eventually performs dozens of human jobs, the revenue opportunity becomes enormous. That gives the company a much larger option value.

Unitree sits between those two extremes because it already sells thousands of robots while also participating in the humanoid narrative.

Can Physical AI software companies become more valuable than robot makers?

Yes. Physical AI software companies can become more valuable than many hardware manufacturers if their intelligence works across enough machines and customers.

Wayve is already worth $8.6 billion without trying to manufacture its own global fleet of cars. Its strategy is to license an embodied-AI driving system across vehicle brands. The company raised $1.2 billion in its Series D and secured additional capital tied to deployment, while automakers including Mercedes-Benz, Nissan and Stellantis participated in the ecosystem around the round.

Skild AI pushes the idea further. Its ambition is one general-purpose intelligence layer that controls many kinds of robots. A model that genuinely works across warehouse robots, arms, humanoids and other machines could be sold far more broadly than a single piece of hardware.

Generalist AI is pursuing the same basic opportunity. Its foundation model is designed to learn new tasks and transfer across different robot configurations.

The economics are attractive in theory. The manufacturer pays for motors, batteries, metal, assembly and logistics every time it ships another robot. A software company can potentially deploy the same model repeatedly at much higher incremental margins.

The catch is technical: robot intelligence does not run independently from hardware. Sensors, hands, joints, safety systems and control loops differ across machines.

Which Physical AI valuations look the most fragile?

The most fragile Physical AI valuations are generally the ones where the current number depends on an active financing, a proxy or a broad estimate rather than a recently completed transaction.

Anduril illustrates the distinction. Its completed $61 billion financing provides a strong anchor. A value around $100 billion may ultimately be confirmed, but until another deal closes there, the higher figure carries more uncertainty.

The same principle applies to companies such as Physical Intelligence, Applied Intuition, Genesis AI, 1X and several younger Chinese humanoid startups in our ranking. Their current ranges may be reasonable, but we should not treat the midpoint with the same confidence as Waymo's $126 billion, Figure's $39 billion, XPENG Robotics' $6.3 billion-plus or Saronic's $9.25 billion.

Market narratives add another layer of fragility.

Humanoids are attracting unusually high expectations. Robot foundation models are being valued before the industry knows whether one intelligence layer can truly work across many robot bodies. Defense valuations assume elevated procurement and geopolitical demand remain durable.

The key is simple: an estimate should still look like an estimate, not like a completed transaction.

What separates a $1 billion Physical AI startup from a $10 billion one?

A $10 billion Physical AI company usually combines a huge potential market with evidence that it could own an important layer of that market.

Pudu has real deployments and substantial shipments but remains around $1.5 billion because investors see a relatively understandable commercial-robotics business.

Skild AI exceeds $14 billion because investors believe one robot intelligence system could spread across many manufacturers and industries.

Figure reached $39 billion because investors are pricing both humanoid hardware and a general-purpose AI platform.

Waymo reaches $126 billion because it combines advanced autonomy, operating fleets, a giant transportation market and years of technical data.

Defense companies follow the same pattern. A drone startup can reach several hundred million dollars. A company that could become an autonomous defense prime across air, sea, ground and software can reach tens of billions.

Is the Physical AI market already in a valuation bubble?

Parts of Physical AI clearly show bubble-like valuation behavior, but the market is too uneven to describe the entire category that way.

The strongest warning comes from valuation multiples. Unitree traded above 150 times expected sales around its early public-market surge according to Financial Times analysis. Figure's $39 billion valuation arrived while the commercial humanoid market remained tiny. Several young robot-intelligence companies have reached multibillion-dollar valuations before developing mature revenue bases.

Defense has produced similar concerns. Financial Times reporting recently highlighted debate around European defense startup valuations, including Helsing at roughly 32 times projected annual revenue.

At the same time, some Physical AI companies already have unusual strategic value.

Waymo operates real autonomous transportation. Pudu has shipped more than 120,000 robots. Defense-autonomy companies are winning government contracts. Quantum Systems says its valuation growth has come alongside rapidly growing revenue and profitability. Skild AI reports more than $100 million of ARR.

Physical AI currently contains real commercial businesses and very optimistic future-platform bets at the same time. Some of those bets will look cheap if the market develops as investors expect. Others will not.

So which Physical AI startups are really winning the valuation race?

Waymo is the clear valuation leader in Physical AI today, while Anduril, Figure and Unitree form a second tier whose valuations reflect very different combinations of commercial proof and future expectations.

Waymo's $126 billion value is supported by a completed financing and a real robotaxi business. It is the cleanest number at the top.

Anduril is the biggest caveat in the ranking. The company clearly belongs among the largest Physical AI businesses, but we would keep a distinction between its roughly $61 billion completed-round anchor and estimates approaching $100 billion.

Figure owns the strongest private humanoid valuation with a confirmed $39 billion post-money round. Unitree has demonstrated that public investors can value a robotics manufacturer at tens of billions of dollars, although its share price has already moved dramatically from its IPO peak.

Below them, three groups are moving quickly.

Defense-autonomy companies such as Shield AI, Saronic, Quantum Systems and TEKEVER are gaining value alongside expanding military demand. Robot-intelligence companies such as Skild AI, Physical Intelligence and Generalist are attracting software-style expectations. Humanoid manufacturers including Apptronik, NEURA and XPENG Robotics are trying to turn large financing rounds into repeatable industrial deployment.

Our dataset contains almost 100 meaningful companies, roughly 60 around unicorn territory and more than half a trillion dollars of estimated combined value. Around 70% sits in the ten largest companies.

The biggest bets are now concentrated in four places: autonomous transportation that already works, defense autonomy with urgent customers, humanoids that could automate human environments, and general-purpose robot intelligence that could become the software layer underneath many machines.