What are the fundraising trends in the embodied AI market?
In our updated market reports, you will find everything you need
SUMMARY
This report analyzes publicly disclosed equity funding rounds raised by pure-play embodied AI companies across full-year 2024, full-year 2025, and year-to-date 2026 through July. It only includes disclosed rounds of $300K or more and focuses on companies built around robotic foundation models, humanoid robot intelligence, VLA/action models, robot learning, control policies, simulation-to-real systems, embodied datasets, or real-world embodied AI deployment.
The embodied AI market is accelerating sharply. Full-year funding rose from about $3.09B in 2024 to about $4.23B in 2025, and year-to-date 2026 had already reached about $5.60B by early July. That means 2026 funding had already exceeded the entire 2025 total before the year was half over.
The latest increase is driven much more by larger rounds than by more deals. Year-to-date 2026 had 14 disclosed deals versus 10 over the comparable 2025 period, but capital rose from about $835M to about $5.60B. The median round jumped from about $39M to about $238M.
The embodied AI market is financially maturing faster than it is technically de-risking. A market with a $238M median round and 12 of 14 year-to-date 2026 deals above $100M is no longer behaving like a normal early-stage robotics category, even though broad commercial deployment is still not fully proven.
Capital is moving strongly toward follow-on companies and perceived platform winners. In year-to-date 2026, follow-on deals represented 12 of 14 deals and captured about 86% of capital. First financings still happened, but they accounted for only 14% of deals and about 14% of capital.
Robotic Foundation Models are now the leading capital category. They captured about $2.28B in year-to-date 2026, or roughly 41% of the total, ahead of Humanoid Robot Software at about $1.92B. The investable thesis is therefore broader than humanoid bodies alone.
Humanoid Robot Software remains the most powerful capital magnet when investors believe the company owns a full stack. The category had only 2 year-to-date 2026 deals but captured about 34% of capital, driven by Apptronik and NEURA Robotics.
The market is globalizing by deal count but remains concentrated by capital. Asia-Pacific produced half of year-to-date 2026 deals, North America produced about 43%, and Europe produced one deal. But North America still captured about 56% of capital, while Europe’s entire 25% share came from NEURA Robotics alone.
Investor participation is broadening quickly. Year-to-date 2026 had about 108 disclosed investors and about 38 unique tier-1 investors, already above the full-year 2025 disclosed investor count. The most important investors are not only venture funds but also compute, cloud, automotive, industrial, internet-platform, and robotics supply-chain strategics.
The core market interpretation is that embodied AI has entered a platform-capitalization race. Investors are no longer asking whether embodied AI is fundable; they are asking which companies can control the loop between model, robot body, deployment environment, and proprietary data.
Is more or less capital going into the embodied AI market?
More capital is going into the embodied AI market, and the acceleration is now unmistakable. The clearest fresh comparison is year-to-date 2026 versus the same period in 2025: funding reached about $5.60B through early July 2026, compared with about $835M over the comparable 2025 period.
That is roughly a 6.7x increase in capital in the freshest like-for-like period. The fuller comparison also points upward: full-year 2025 reached about $4.23B, up from about $3.09B in full-year 2024. So both the latest comparison and the most recent full-year comparison point in the same direction.
The important interpretation is that the embodied AI market is not just drifting upward. Year-to-date 2026 had already exceeded full-year 2025 by about $1.38B, even though 2026 was still incomplete. That turns the market from a large frontier category into an aggressively capitalized platform race.
The caveat is concentration. In year-to-date 2026, the top 3 deals captured 59.3% of all capital, and the top 5 captured 74.4%. So more capital is entering the embodied AI market, but it is not becoming broadly available to every company with a robotics or humanoid story.
The practical takeaway is that capital availability in the embodied AI market is highly conditional. Investors are funding companies that can plausibly claim robotic foundation models, humanoid deployment platforms, VLA/control systems, industrial data loops, strategic deployment access, or some combination of those advantages.
Is embodied AI funding activity driven by more deals or larger rounds?
Embodied AI funding activity is being driven much more by larger rounds than by more deals. Year-to-date 2026 had 14 deals versus 10 over the comparable 2025 period, so deal count increased by 40%, while capital increased roughly 6.7x from about $835M to about $5.60B.
The average round size confirms the point. The average year-to-date 2026 round was about $400M, compared with about $84M over the comparable 2025 period. The median round size rose from about $38.5M to about $238M, which means the increase is not only one huge outlier pulling up the average.
The full-year comparison adds nuance. Full-year 2025 had 25 deals versus 17 in 2024, so the market did broaden by deal count. Total capital also rose from about $3.09B to about $4.23B. But average deal size slipped slightly, from about $182M in 2024 to about $169M in 2025, while median round size rose from $70M to $100M.
The cleaner interpretation is that 2025 broadened the embodied AI market by adding more companies and more mid-to-large rounds. Then 2026 changed the structure of the market by pushing the typical disclosed round into mega-round territory.
So the embodied AI market has seen both more deals and larger rounds, but the current acceleration is overwhelmingly a larger-round story. The number of funded companies is increasing, but the capital curve is steepening much faster than the company count.
Is embodied AI capital moving toward later-stage or earlier-stage companies?
Embodied AI capital is moving toward later-stage companies, especially in the freshest 2026 period. In year-to-date 2026, late-stage capital, defined as Series B and later plus Growth Equity, reached about $4.12B, or 73.5% of total funding.
That is a sharp change from the comparable 2025 period. Through early July 2025, early and unknown-stage rounds represented about 85% of capital, while late-stage Series B+ capital represented only about 15%. The 2025 market still looked like a formation market; the 2026 market looks like a scale-up market.
The round-level evidence is clear. Skild AI, NEURA Robotics, Apptronik, Generalist AI, X Square Robot, AI2 Robotics, ROBOTERA, and Pudu Robotics all point toward investors backing companies that already look like platform contenders, not just early experiments.
The full-year comparison is somewhat mixed but still useful. In 2024, late-stage capital was about 65% of total capital. In 2025, late-stage capital was about 55%, because large Series A and Seed rounds broadened the market. But year-to-date 2026 swung strongly back toward later-stage and growth-style capitalization.
The key nuance is that stage labels are unusually weak in embodied AI. A Series A can be $32M, $320M, $450M, or $520M. So the better interpretation is not simply that later-stage companies are winning; it is that investors are funding companies that already behave like platform-scale companies, even when the official stage label looks early.
Is the embodied AI market maturing or still experimental?
The embodied AI market is maturing financially, but it is still experimental technically. Financially, year-to-date 2026 does not look like a scattered early-stage category: the median round was about $238M, the average round was about $400M, and 12 of 14 deals were above $100M.
Technically, the market is still experimental because the largest funding claims are built around future generalization, closed-loop learning, cross-embodiment control, robot data flywheels, and scalable real-world deployment. Those are powerful claims, but they are not the same thing as broad, proven autonomous labor substitution.
The 2024 to 2025 comparison shows the transition. In 2024, funding was dominated by a few large validation rounds such as Wayve, Figure AI, Physical Intelligence, Skild AI, and Collaborative Robotics. In 2025, the market broadened to 25 deals across more categories, including VLA models, simulation training, robot learning platforms, and real-world deployment companies.
By year-to-date 2026, the embodied AI market had become even more capital-intensive. Robotic Foundation Models alone captured about $2.28B, and Humanoid Robot Software captured about $1.92B. Investors are increasingly treating embodied AI as an infrastructure race rather than as a conventional robotics product cycle.
The best description is that the embodied AI market is an early technical market with late-stage financial behavior. Investor conviction has matured faster than commercial proof, which creates both opportunity and overcapitalization risk.
Are new startups still entering the embodied AI market?
New startups are still entering the embodied AI market, but the share of capital going to new entrants is falling. In year-to-date 2026, first financings represented only 2 of 14 deals, or 14.3%, and about $770M of $5.60B, or 13.7% of capital.
That is much lower than the comparable 2025 period. Through early July 2025, first financings represented 5 of 10 deals, or 50%, and about 26% of capital. So the market has become much less centered on brand-new company formation.
The full-year trend points in the same direction. In 2024, first financings were about 41% of deals and about 18% of capital. In 2025, first financings were 36% of deals but only about 8% of capital. New companies were still being formed, but the money was already moving toward follow-on companies.
This does not mean the embodied AI market is closed to new startups. Rhoda AI and General Intuition show that a new or newly surfaced company can still raise a very large round if it has frontier-level technical credibility, founder quality, or strategic investor support.
The real rule is that ordinary new entrants now face a much higher bar. A new embodied AI company needs a sharp wedge: proprietary interaction data, a credible robot foundation model, a unique action model, deep industrial access, or a team that investors believe can compete in frontier AI.
Are more investors entering the embodied AI market?
More investors are entering the embodied AI market, and the clearest signal is the year-to-date 2026 investor count. Through early July 2026, the market had about 108 disclosed investors and about 38 unique tier-1 investors, already above the roughly 82 disclosed investors in full-year 2025.
The comparison with the same period in 2025 is even more striking. The comparable 2025 period had about 39 disclosed investors and about 18 unique tier-1 investors. So the embodied AI market is attracting more capital providers, not just larger checks from the same narrow group.
The investor mix also matters. The market now includes frontier-AI investors, deep-tech funds, sovereign and institutional capital, chip companies, cloud platforms, automotive groups, industrial companies, logistics players, internet platforms, and robotics supply-chain strategics.
Strategic investors such as NVIDIA, Amazon, Google, Qualcomm, Mercedes-Benz, John Deere, Bosch, Schaeffler, Xiaomi, Alibaba, ByteDance, Meituan, SF Group, and Baidu are especially important because they can bring compute, manufacturing, distribution, deployment sites, pilots, or data access. In embodied AI, those non-cash resources can matter as much as the funding itself.
The better interpretation is that investor participation is broadening, while serious capital remains concentrated. More investors want exposure to the embodied AI market, but the largest rounds are still going to a relatively small group of perceived platform leaders.
Are top investors getting more or less active in embodied AI?
Top investors are getting more active in the embodied AI market, but they are spreading their bets across competing technical theses rather than converging on one obvious winner. In year-to-date 2026, multiple investors or investor groups appeared in more than one qualifying deal, including Bezos Expeditions, ByteDance, HongShan / Sequoia China, NVentures, Khosla Ventures, Horizon Investment, and Tsinghua Holding Tiancheng.
That is a stronger repeat-investor signal than the comparable 2025 period, when Khosla Ventures was the only clearly disclosed investor appearing in more than one accepted deal. It also continues the pattern from full-year 2025, when NVentures, Khosla, First Round, CRV, Google/Alphabet-linked capital, Intel Capital, Salesforce, and LG Technology Ventures appeared repeatedly.
Looking back to 2024, repeat investors included OpenAI, Khosla, Lux, Jeff Bezos / Bezos Expeditions, Sequoia, NVIDIA, Microsoft, SoftBank, Thrive, General Catalyst, Matrix Partners China, BlueRun Ventures, and IDG Capital. That continuity matters because it suggests serious investors are not simply chasing one news cycle.
The main shift is portfolio construction. Top investors are not only backing humanoids or only backing robot foundation models. They are placing bets across robot brains, humanoid platforms, industrial autonomy, VLA systems, control policies, and real-world deployment companies.
So top investors are more active, but their activity is still exploratory. The strongest investors appear to have category-level conviction without certainty about which architecture, embodiment, or deployment pathway will ultimately dominate.
Which embodied AI subcategories are gaining momentum?
Robotic Foundation Models are gaining the most momentum in the embodied AI market. Full-year 2025 robotic foundation model companies raised about $745M across 5 deals, while year-to-date 2026 had already reached about $2.28B across 5 deals.
That means the category raised more than 3 times its full-year 2025 capital total before 2026 was complete. The category also became the largest year-to-date 2026 funding bucket, with about 41% of total capital.
Vision Language Action Models are also gaining momentum. VLA models went from $27.8M in full-year 2024 to $626M in full-year 2025, then reached about $421M in year-to-date 2026. The category has fewer deals than humanoids or foundation models, but its deal sizes now show that investors treat action-model intelligence as strategically important.
Humanoid Robot Software is gaining capital momentum, although not deal-count momentum. The category raised about $1.06B in 2024, about $2.16B in 2025, and already about $1.92B in year-to-date 2026. The 2026 capital came from only 2 deals, Apptronik and NEURA, which means momentum is concentrated in a few full-stack companies.
Robot Control Policies are also gaining attention. Rhoda AI’s $450M year-to-date 2026 round is a major step up from the category’s $95M in full-year 2025 and $10.5M in 2024. Robot Learning Platforms are emerging more slowly, with Trener’s $32M round showing that the tooling layer is fundable but not yet capitalized like the full-stack platform layer.
Which embodied AI subcategories are losing momentum?
Real World Robot Deployment is losing relative momentum in the embodied AI market, even though absolute funding remains meaningful. The category captured about $1.22B in full-year 2024, fell to about $551M in full-year 2025, and reached about $496M in year-to-date 2026.
The year-to-date 2026 number is already close to the full-year 2025 total, so the category is not disappearing. But its share of year-to-date 2026 capital was only about 9%, far below Robotic Foundation Models and Humanoid Robot Software.
The practical reading is that real-world deployment is a validation signal but not always a valuation premium. ROBOTERA and Pudu Robotics show that investors value deployment, but the market is assigning larger checks to companies that claim scalable intelligence layers rather than only commercial robot rollout.
Embodied AI Datasets remain absent as standalone funding categories across the evidence reviewed. That absence is important because every large embodied AI company needs data. The market appears to be internalizing dataset value inside foundation-model, humanoid, and deployment companies rather than financing standalone dataset suppliers.
Simulation Training Environments also remain underrepresented. Flexion Robotics’ $50M 2025 round was meaningful, but there was no comparable accepted year-to-date 2026 round. Simulation is clearly important to embodied AI, but investors seem more likely to fund simulation when it is embedded inside a broader robot intelligence company.
Which regions are gaining momentum in embodied AI funding?
Asia-Pacific is gaining the most deal-count momentum in the embodied AI market, while North America is gaining the most capital-scale momentum. In year-to-date 2026, Asia-Pacific accounted for 7 of 14 deals, or 50% of deal count, compared with just 1 of 10 deals over the comparable 2025 period.
Asia-Pacific capital also rose sharply, from about $15M over the comparable 2025 period to about $1.08B in year-to-date 2026. That reflects a clear step-up in Chinese and Asia-Pacific embodied AI activity, including X Square Robot, AI2 Robotics, RLWRLD, ROBOTERA, and Pudu Robotics.
North America remains the strongest region by capital. Year-to-date 2026 North America raised about $3.12B, compared with about $671M over the comparable 2025 period. Even though North American deal count moved from 7 comparable-period deals in 2025 to 6 in year-to-date 2026, capital rose sharply because average deal size exploded.
Europe gained capital momentum only through one very large round. NEURA Robotics’ up-to-$1.4B Series C gave Europe 25% of year-to-date 2026 capital, but Europe had only 1 accepted deal in that window. That is a strong flagship signal, not broad regional depth.
The best interpretation is that Asia-Pacific is gaining breadth, North America is gaining scale, and Europe is gaining visibility through a single standout company. The embodied AI market is more global in company formation than it is in capital distribution.
Which regions are losing momentum in embodied AI funding?
Europe is losing momentum by breadth, even though it gained by capital in year-to-date 2026. Full-year 2025 Europe had 4 accepted embodied AI deals and about $202M in capital, while year-to-date 2026 had only 1 European deal.
That one European deal was enormous, so Europe looks strong in dollars. But the regional signal is fragile because NEURA Robotics accounts for the entire European year-to-date 2026 funding total. Without NEURA, Europe disappears from the disclosed 2026 picture.
The Middle East also lost visible momentum after Mentee Robotics’ $17M 2024 round. There were no accepted Middle East deals in full-year 2025 or year-to-date 2026. That does not mean there is no regional robotics or AI activity, but it does mean the public pure-play embodied AI funding evidence is not showing sustained depth there.
Latin America and Africa remain absent from the accepted disclosed funding evidence across the periods reviewed. That absence should not be read as a lack of talent or use cases, but it does show that large disclosed embodied AI funding rounds remain concentrated in North America, Asia-Pacific, and a small number of European companies.
North America is not losing momentum in capital, but it is losing relative deal-count dominance. North America produced 70% of comparable-period 2025 deals and 42.9% of year-to-date 2026 deals. That is not North American weakness; it reflects Asia-Pacific catching up in disclosed activity.
Is the embodied AI market becoming more global or more regionally concentrated?
The embodied AI market is becoming more global by deal count, but it remains regionally concentrated by capital. In year-to-date 2026, Asia-Pacific produced 50% of deals, North America produced about 43%, and Europe produced about 7%.
That is more globally distributed than the comparable 2025 period, when North America produced 70% of deals. The deal-count story therefore points to geographic broadening, especially because China and Asia-Pacific companies became much more active in 2026.
By capital, however, the embodied AI market remains concentrated. North America captured about 56% of year-to-date 2026 capital, Europe captured 25%, and Asia-Pacific captured 19%. Europe’s entire share came from NEURA Robotics, so Europe’s capital strength does not yet represent a broad base of companies.
The full-year history reinforces the point. In 2025, North America captured about 66% of capital, Asia-Pacific about 29%, and Europe about 5%. In 2024, North America captured 50%, Europe 37%, Asia-Pacific 12%, and the Middle East less than 1%.
So the embodied AI market is globalizing in participation but not democratizing in capital access. Large capital pools still cluster around North America, China/Asia-Pacific, and a few exceptional European companies.
Is embodied AI capital moving toward proven winners or new opportunities?
Embodied AI capital is moving strongly toward proven winners, while still leaving room for selective new opportunities. In year-to-date 2026, follow-on deals represented 12 of 14 deals and captured about 86% of total capital.
The comparable 2025 period was much more open to new opportunities. Through early July 2025, first financings represented 5 of 10 deals and about 26% of capital. Full-year 2025 still had 9 first financings, but those first financings captured only about 8% of total capital, showing that the shift toward follow-on winners was already underway.
The 2026 market intensified that split. Skild AI, Apptronik, NEURA Robotics, X Square Robot, ROBOTERA, Generalist AI, AI2 Robotics, RLWRLD, and Pudu Robotics are all follow-on stories. These companies are not merely entering the category; they are being recapitalized as possible winners.
The new-opportunity capital is concentrated in unusually ambitious cases such as Rhoda AI and General Intuition. Those companies show that investors will still fund new entrants, but only when the technical scope looks large enough to compete with existing platform contenders.
The better interpretation is that investors are no longer asking whether embodied AI is a fundable theme. They are asking which companies can accumulate the models, robot data, deployment channels, strategic relationships, and engineering teams needed to survive the scaling race.
Is the embodied AI market becoming winner-takes-most?
Yes, the embodied AI market is becoming winner-takes-most by capital allocation, though not winner-takes-one. In year-to-date 2026, the top 3 rounds captured 59.3% of capital, the top 5 captured 74.4%, and the top 10 captured 93.9%.
That concentration is very high. The bottom half of year-to-date 2026 deals captured only 14.9% of capital, which means most funded companies are not shaping the market’s headline funding narrative.
The concentration pattern has moved over time. Through early July 2024, the top 3 deals captured 86% of capital. Full-year 2024 top 3 deals captured about 69%. Full-year 2025 top 3 deals captured about 47%, suggesting the market briefly broadened. Year-to-date 2026 then re-concentrated around very large platform rounds.
The important nuance is that 2026 is not a single-company story. Skild AI and NEURA Robotics each reached about $1.4B, and Apptronik, Rhoda AI, Generalist AI, X Square Robot, ROBOTERA, Pudu Robotics, and AI2 Robotics all raised very large rounds.
So the embodied AI market is not becoming winner-takes-one. It is becoming winner-takes-most across a small group of platform contenders that are likely to shape talent markets, customer expectations, supplier relationships, and investor benchmarks.
Is the next wave of embodied AI winners becoming visible?
The next wave of embodied AI winners is becoming visible, but the final winners are not yet settled. The clearest candidates are companies that combine large capital raises with at least one of three signals: foundation-model ambition, real-world deployment access, and strategic investor support.
The visible platform group includes Skild AI, NEURA Robotics, Apptronik, Generalist AI, X Square Robot, ROBOTERA, Rhoda AI, AI2 Robotics, Physical Intelligence, Figure AI, FieldAI, Dyna Robotics, Galbot, and potentially Pudu Robotics. These companies stand out because their funding is tied to robot brains, humanoid fleets, VLA models, industrial deployments, physical AI platforms, or closed-loop learning systems.
The repeat-funding pattern is especially important. X Square Robot raised multiple rounds in 2026. ROBOTERA raised multiple large 2026 rounds. Apptronik followed its 2025 Series A with a major 2026 extension. Skild AI followed its 2024 Series A with a 2026 mega-round.
That repeat funding suggests investors are not only responding to demos or narratives. They are backing ongoing platform races where the same companies keep attracting larger pools of capital.
But funding visibility is not the same as commercial victory. The strongest future winners will need to turn capital into repeat deployments, useful autonomy, data flywheels, manufacturing reliability, customer ROI, and measurable improvement in robot capability over time.
Is the embodied AI funding landscape fragmenting or consolidating?
The embodied AI funding landscape is consolidating by capital and fragmenting by technical approach. By capital, year-to-date 2026 is clearly consolidating around fewer, larger rounds: 12 of 14 deals were above $100M, and the top 5 represented 74.4% of capital.
By technical approach, the market is still fragmented. Funding is flowing into robotic foundation models, humanoid robot software, real-world robot deployment, robot control policies, VLA models, and robot learning platforms. Investors are not yet converging on one architecture, body type, or deployment pathway.
The 2024 market was shaped by large rounds in Real World Robot Deployment, Humanoid Robot Software, and Robotic Foundation Models. The 2025 market broadened into VLA, simulation, and robot learning infrastructure. The year-to-date 2026 market intensified foundation models and humanoids while keeping meaningful capital in VLA, control, deployment, and tooling.
The best reading is that funding is consolidating around leading companies while the technical map remains open. The embodied AI market is not yet a settled category with one dominant architecture. It is a capital-intensive contest among several possible control points.
Where is investor attention shifting in embodied AI?
Investor attention in the embodied AI market is shifting toward robot foundation models, cross-embodiment intelligence, humanoid deployment platforms, and closed-loop learning systems. The strongest evidence is that Robotic Foundation Models captured about $2.28B in year-to-date 2026, the largest category by capital.
Humanoid Robot Software also captured about $1.92B from only 2 deals, which shows that investors still care deeply about humanoid systems when the company can credibly claim a full-stack platform. Apptronik and NEURA Robotics are not being funded as simple hardware companies; they are being funded as embodied AI deployment platforms.
Investor attention is also shifting toward data flywheels. NEURA emphasizes physical AI platforms and robot training infrastructure. Apptronik is tied to Apollo humanoid deployment. X Square Robot emphasizes physical AI foundation models and VLA capabilities. RLWRLD emphasizes training robotics foundation models in live industrial environments. Rhoda AI emphasizes closed-loop video-predictive robot control.
The shift is not simply from hardware to software. It is from robot products to learning systems. Investors appear to care less about the robot body as an isolated machine and more about whether a company can continuously improve through data, simulation, deployment feedback, and cross-task generalization.
The strongest investor-attention rule is that the embodied AI market rewards companies that can plausibly own the loop between model, body, environment, and data. Companies that only own one piece of the stack can still be funded, but they need a very strong wedge to compete with full-stack or foundation-model players.
INSIGHTS
The insights below come from reviewing publicly disclosed equity funding rounds in the embodied AI market across full-year 2024, full-year 2025, and year-to-date 2026 through July.
- The embodied AI market has crossed from venture formation into capitalization warfare. Year-to-date 2026 capital of about $5.60B already exceeded full-year 2025 funding, which means the financial pace of the market has outrun the normal startup adoption cycle.
- The most important change from 2025 to 2026 is not deal count; it is the repricing of credible embodied AI companies. Deal count rose 40% over the comparable period, while capital rose about 6.7x, so the current acceleration is overwhelmingly a round-size story.
- The median round is now more informative than the average round. A year-to-date 2026 median of $238M means large financings are no longer limited to one or two outliers; the middle of the disclosed funding market has moved into mega-round territory.
- The embodied AI market is behaving more like frontier AI infrastructure than traditional robotics. A conventional robotics market would not normally have 12 of 14 year-to-date deals above $100M before broad deployment maturity.
- Foundation-model language has become one of the strongest funding accelerants in embodied AI. Robotic Foundation Models captured 40.8% of year-to-date 2026 capital, more than any other category, even though humanoids remain the more visible public narrative.
- Humanoids remain the clearest capital magnet when investors believe the company owns more than the physical body. Apptronik and NEURA Robotics raised enormous rounds because they are framed as embodied AI platforms, not simply robot manufacturers.
- Real-world deployment is necessary evidence, but it is not automatically the highest-valued layer. Real World Robot Deployment had 21.4% of year-to-date 2026 deals but only 8.9% of capital, suggesting deployments validate the market but do not always create frontier-AI valuation multiples.
- The market is rewarding generalization claims more than narrow productivity claims. Companies promising cross-body, cross-task, or general physical intelligence are receiving larger checks than companies focused only on single-environment automation.
- Category labels are analytically useful but strategically incomplete. The strongest embodied AI companies increasingly blend model, body, deployment, data, and training infrastructure, so a company’s category often understates its actual ambition.
- Embodied AI Datasets remain conspicuously absent as a standalone funding category. This suggests data value is being internalized by full-stack robot and model companies rather than financed as an independent supplier layer.
- Robot-learning infrastructure is undercapitalized relative to its strategic importance. Trener’s $32M round is meaningful, but tiny compared with the billions going into companies that will ultimately depend on training infrastructure, data workflows, and robot-learning pipelines.
- The market has become less friendly to ordinary first financings. First financings fell from 50% of comparable-period 2025 deals to 14.3% of year-to-date 2026 deals, which means investors now prefer already validated teams or unusually credible new platforms.
- A new embodied AI startup can still raise major capital, but only if it looks like a platform from day one. Rhoda AI and General Intuition show that new entrants are not excluded, but the bar is frontier-AI credibility rather than standard seed-stage traction.
- Stage labels are becoming misleading. Series A can mean a $32M robot-skills platform, a $320M action-model company, a $450M robot-control company, or a $520M humanoid extension, so stage alone is a weak proxy for maturity.
- The embodied AI market is consolidating financially before it has consolidated technically. Capital is flowing to a small number of companies, but investors are still backing multiple architectures, including humanoids, VLAs, robot brains, industrial control policies, and deployment platforms.
- The geographic pattern is bifurcated: Asia-Pacific is gaining breadth, while North America is gaining capital scale. Asia-Pacific produced 50% of year-to-date 2026 deals, while North America captured 55.7% of capital.
- Europe’s signal is powerful but fragile. NEURA Robotics gives Europe 25% of year-to-date 2026 capital, but Europe had only one accepted deal, so the region’s apparent strength depends on a single company.
- China’s embodied AI market appears more deployment-and-ecosystem driven than the U.S. market. Chinese deals more often include industrial, internet-platform, logistics, automotive, or state-linked investors, which points to commercialization through existing industrial systems.
- The U.S. market appears more model-and-platform driven. Large North American rounds are more often framed around robot foundation models, robot brains, action models, or generalizable physical intelligence.
- Strategic investors are becoming more important than purely financial investors. NVIDIA, Google, Amazon, Mercedes-Benz, John Deere, Bosch, Schaeffler, Xiaomi, Alibaba, ByteDance, Meituan, Baidu, SF Group, and Qualcomm matter because they can influence compute, supply chain, data access, pilots, or distribution.
- Repeat investors are spreading bets across the stack rather than choosing one winner. Khosla, NVentures, Bezos-related capital, HongShan, ByteDance, Horizon, and Tsinghua-linked capital appear across different embodied AI approaches, which suggests category conviction is stronger than certainty about the final architecture.
- The next wave of winners is visible but not guaranteed. Large rounds identify likely contenders, but the decisive proof will be repeat deployments, utilization, autonomy quality, unit economics, and data accumulation.
- The most useful evaluation rule is whether a company owns a defensible data loop. Companies with model claims but no proprietary data source, deployment environment, or embodiment strategy should be discounted.
OUR METHODOLOGY TO BUILD THIS TRACKER
We built this embodied AI funding tracker by reviewing publicly disclosed equity rounds raised by pure-play embodied AI companies across full-year 2024, full-year 2025, and year-to-date 2026 through July. A company counts as pure-play when more than 80% of its activity is dedicated to embodied intelligence, robot foundation models, humanoid robot intelligence, VLA/action models, robot control policies, robot learning platforms, simulation-to-real training, embodied datasets, or real-world embodied AI deployment.
We define the embodied AI market as AI models and systems that learn through physical interaction, simulation, sensors, and robotic bodies. It includes Robotic Foundation Models, Humanoid Robot Software, Simulation Training Environments, Vision Language Action Models, Robot Control Policies, Embodied AI Datasets, Robot Learning Platforms, and Real World Robot Deployment. It excludes broader adjacent markets unless the product is built specifically for this use case.
We applied four core filters. First, we only included equity rounds, so grants, debt-only financings, acquisitions, SPAC transactions, structured credit, and business combinations were excluded. Second, we only counted rounds of $300K or more. Third, we kept only pure-play companies where embodied AI or physical intelligence is central to the business. Fourth, every included entry had to be supported by a direct company announcement, press release, tier-1 media report, specialized robotics publication, or relevant regional publication.
We excluded broader robotics, automation, autonomous vehicle, drone, computer vision, chip, and generic AI companies unless the company’s product was explicitly built around embodied intelligence, robot learning, humanoid robot intelligence, robotic foundation models, VLA/action systems, or physical-world AI deployment. Borderline cases were only included when the source itself tied the financing to embodied AI or physical AI as a core product thesis.
Undisclosed-amount rounds were excluded from dollar-based metrics because including them would distort totals, averages, medians, concentration ratios, and category shares. When a round was publicly identified but the dollar amount was not disclosed, it was noted separately only when relevant. All metrics are calculated on disclosed-size public rounds, which means privately raised or database-only rounds that were never publicly announced are necessarily missing.
Related blog posts
- Who has raised the most money in the embodied AI market?
- Who are the most valuable companies in the embodied AI market?
- How active is venture funding in the embodied AI market right now?
