Robotics funding from mid-September to mid-October 2026: chips, data and safety took the private money while humanoid makers turned to stock markets

In the 30 days to October 14, 2026, MEKAWIRE verified 14 private robotics rounds with disclosed amounts worth $1,071.3 million. Compute chips took 51.3%, robot models 20.5% and data, perception and safety software 8.7%, while humanoid makers received $10 million, or 0.9%, against 13.4% between January and mid-September.
The four October rounds, Mecka, Multiply Labs, SafeWorld and Minerva Humanoids, paid for what stands between a capable robot and a paid deployment, while humanoid makers raised through an IPO, a SPAC and a Form D offering.
Humanoid makers received $10 million of the $1,071.3 million that private robotics rounds raised between September 15 and October 14, 2026, according to MEKAWIRE's verified funding records. That single round was Minerva Humanoids' pre-seed, and it pays for a robot that a person drives through a VR headset. Robot compute chips took $550 million, a robot-model lab took $220 million, and the four rounds announced so far in October went to a data supplier, a drug-manufacturing robot, a safety simulator and that teleoperated humanoid.
The split matters because it shows where investors now place the bottleneck. Between January 1 and September 14, humanoid makers took 13.4% of the disclosed money in our 2026 records and robot-model developers 30.4%. In the latest 30 days the money moved to the layers that decide whether a robot gets paid work at all: the chip on board, the data it learns from, the evidence that it is safe near people and a production process a customer can validate. Humanoid makers did not stop raising. They moved to the exchanges. This piece sets out the numbers, the investors behind them and what would prove the reading wrong.
The count covers 14 disclosed rounds in a 30-day window
We took every non-demo funding row in the MEKAWIRE database with a disclosed US dollar amount and a primary source. Form D and Regulation A partial sales and IPOs are left out because they are filings or listings, not priced private rounds. The window from September 15 to October 14 holds 14 such rounds worth $1,071.3 million, and January 1 to September 14 holds 36 rounds worth $10,350.5 million. Six financings in the window carry no dollar figure, among them Qibitech's Series B of about ¥600 million, Extend Robotics' GBP 2.6 million pre-A and the Hong Kong listings of Camsense and Direct Drive Tech.
Our coverage is not the whole market, and the months before October were filled partly by backfill that favoured large rounds. The median round in the window is $18.3 million against $158.5 million earlier in the year, but that gap says as much about our coverage as about the market, so the comparison below uses shares of money rather than counts or medians. Even on shares, the change is large.
- 2026-10-07
- Mecka
- Not recorded
- Data, perception, safety software
- Series B
- 60
- Sequoia Capital
- 2026-10-06
- Multiply Labs
- US
- Application robots and systems
- Series B
- 75
- Patrick Soon-Shiong (NantWorks)
- 2026-10-06
- Minerva Humanoids
- US
- Humanoid makers and operators
- Pre-seed
- 10
- General Catalyst
- 2026-10-05
- SafeWorld
- US
- Data, perception, safety software
- Seed
- 12.2
- Shine Capital, a16z speedrun
- 2026-09-30
- Inbolt
- FR
- Data, perception, safety software
- Unnamed round
- 12.5
- Shift4Good
- 2026-09-30
- Feather Robotics
- US
- Components and developer hardware
- Launch round
- 7.6
- Gradient
- 2026-09-30
- Aleph Surgical
- Not recorded
- Application robots and systems
- Pre-seed
- 7.5
- Andreessen Horowitz
- 2026-09-30
- Tangent Robotics
- US
- Components and developer hardware
- Pre-seed
- 4.5
- Toyota Ventures, Fly Ventures
- 2026-09-29
- General Intuition
- US
- Robot models
- Unnamed round
- 220
- None named
- 2026-09-29
- Destro AI
- US
- Data, perception, safety software
- Seed
- 8
- Bonfire Ventures, Base10 Partners
- 2026-09-28
- SiMa.ai
- US
- Compute chips
- Series C
- 150
- Amplify, Fidelity Management & Research
- 2026-09-17
- D-Robotics
- CN
- Compute chips
- Series C
- 400
- Not named
- 2026-09-17
- Watney
- US
- Application robots and systems
- Series A
- 80
- Hummingbird Ventures, Valor Atreides AI Fund
- 2026-09-17
- Viabot
- US
- Application robots and systems
- Series A
- 24
- Walden International
Source: MEKAWIRE verified funding rows (non-demo, primary sources), as of 2026-10-15. Figures disclosed, not estimated. Total US$1,071.3m. Not shown because no US$ amount was disclosed: Robotplusplus Series C (Sept 15), Qibitech Series B (about ¥600m), Extend Robotics pre-A (GBP 2.6m), Hello Robot NIH grant, and the Hong Kong listings of Camsense and Direct Drive Tech. MEKAWIRE coverage, not the whole market.
As of Oct 15, 2026
Chips and one model lab took seven dollars in ten
D-Robotics' $400 million Series C, for which the Chinese robot-chip company named no investor, and SiMa.ai's $150 million Series C at a $1.45 billion valuation make up 51.3% of the window. General Intuition's $220 million round, its second in 96 days, adds 20.5%. Between January and mid-September no robot-chip company appeared among our funded rows at all, while three defence-autonomy rounds, Saronic's $1.75 billion, Shield AI's $1.5 billion and Overland AI's $100 million, took 32.4%. Humanoid makers and operators fell from 13.4% to 0.9%. Data, perception, safety and orchestration software rose from 1.4% to 8.7%, and the October rounds show what that money buys.
- Defence autonomy (air, sea, land)
- 3
- 3,350
- 32.4
- 0
- 0
- 0
- Robot models
- 6
- 3,150
- 30.4
- 1
- 220
- 20.5
- Application robots and systems
- 18
- 2,325
- 22.5
- 4
- 186.5
- 17.4
- Humanoid makers and operators
- 7
- 1,383.5
- 13.4
- 1
- 10
- 0.9
- Data, perception, safety and orchestration software
- 2
- 142
- 1.4
- 4
- 92.7
- 8.7
- Compute chips
- 0
- 0
- 0
- 2
- 550
- 51.3
- Components and developer hardware
- 0
- 0
- 0
- 2
- 12.1
- 1.1
- Total
- 36
- 10,350.5
- 100
- 14
- 1071.3
- 100
Share = layer total / period total. Layer = the company's main product in its MEKAWIRE profile, one layer per company. Excludes Form D and Regulation A partial sales (Persona AI, Miso Robotics) and IPOs. Shares rounded to one decimal, so columns may not add to exactly 100. Earlier months were filled partly by backfill that favoured large rounds; compare shares, not counts. Figures disclosed, not estimated.
As of Oct 15, 2026
The October four sell what sits between a model and a paycheque
Mecka closed a $60 million Series B led by Sequoia Capital, the company said on October 7. It records people doing tasks with its own sensor rigs, sells the data and post-trained models, and says it passed a $100 million revenue run rate in June and targets $300 million by year end. Those are company figures and unaudited; taken at face value, the round equals 0.6 times the claimed run rate (60 / 100), a raise sized for expansion rather than survival. Multiply Labs raised a $75 million Series B led by Patrick Soon-Shiong through NantWorks, with AstraZeneca, Lingotto, Teradyne and Strange Ventures as new investors, the company said on October 6. Its robots run GMP production of cell and gene therapies inside drug makers' own plants. Total funding since 2016 now exceeds $100 million, so the Series B is at most three quarters of everything the company has raised in ten years.
SafeWorld left stealth on October 5 with a $12.2 million seed co-led by Shine Capital and a16z speedrun. Its software runs robots through thousands of simulated encounters with people whose motion reacts to the robot, and it says several Fortune 50 companies use it. Minerva Humanoids raised about $10 million led by General Catalyst on October 6 for Roger, a humanoid an operator controls through a VR headset for bomb disposal and energy sites, with first paid pilots this autumn. Each sells a shortcut to paid work: data a model lacks, a process a regulator can validate, safety evidence a plant manager can sign off, and a human operator for jobs autonomy cannot yet do. None of the four is a general-purpose model or a general-purpose humanoid. The investor lists show who wants these layers to exist.
The backers of robot brains are buying the complements
NVIDIA's stake in Mecka is its 11th round in our records, after Figure AI's Series B and C, Generalist AI, NEURA Robotics and Walden Robotics. Qualcomm Ventures and Samsung joined the same round, and Sequoia led it after taking part in Skild AI's $1.4 billion Series C. General Catalyst led Minerva after backing Skild AI's Series A and both General Intuition rounds, and Shine Capital co-led SafeWorld after joining Walden Robotics' $300 million seed. Three of the four October rounds share at least one investor with a robot-model or humanoid company in our records. The exception, Multiply Labs, drew a drug maker and a robot maker, Teradyne, which owns Universal Robots.
The logic is that of complements. A chip or platform supplier gains when training data is plentiful and cheap, and a model investor gains when safety evidence and supervised pilots shorten the road to revenue. That leaves the question of where the humanoid money went.
- Sequoia Capital
- Mecka (lead)
- Skild AI Series C 2026; RobCo Series C 2026; Collaborative Robotics Series A (lead) and B; AMP Series A (lead) to D
- 13
- NVIDIA
- Mecka
- Figure AI Series B and C; Generalist AI 2026; NEURA Robotics Series C 2026; Walden Robotics seed 2026; Skydio, Nuro
- 11
- General Catalyst
- Minerva Humanoids (lead)
- Skild AI Series A; General Intuition Series A and September 2026 round; Standard Bots Series B (lead) and C
- 8
- Qualcomm Ventures
- Mecka
- Figure AI Series C; ANYbotics (lead); Brain Corp Series C and D; Third Wave Automation
- 8
- Innovation Endeavors
- SafeWorld
- Machina Labs Series B (lead); Third Wave Automation Series B and C; Fabric
- 7
- Founders Fund
- Multiply Labs
- OSARO; Built Robotics; Cobalt AI
- 6
- Lux Capital
- Multiply Labs
- Formic Series A (lead); Collaborative Robotics Series B
- 5
- Shine Capital
- SafeWorld (lead)
- Walden Robotics seed 2026
- 2
- Samsung
- Mecka
- THEKER Series A 2026
- 2
- Ora Global
- Multiply Labs
- Inbolt, September 30, 2026
- 2
Counts all non-demo rounds in which the investor is recorded, any year. Investors listed are those in the Mecka, Multiply Labs, SafeWorld and Minerva Humanoids rounds that appear in at least one other round. As of 2026-10-15.
As of Oct 15, 2026
Humanoid makers raised on exchanges and in filings
Unitree Robotics raised RMB6,099.32 million gross when it listed on Shanghai's STAR Market on August 19. Agility Robotics is merging with Churchill Capital Corp XI at a $2.5 billion equity value, with $420.9 million in trust and a $201 million PIPE, and expects to close in the fourth quarter. Persona AI reported $65.36 million sold of a $141.96 million offering in a Form D filed on September 14. None of these is in our private-round totals. Together they show humanoid makers drawing on public investors and exempt offerings, where the price is tested in the open.
- Unitree Robotics
- STAR Market IPO (Shanghai)
- 2026-08-19
- RMB 6,099.32 million gross
- Issue price RMB150.80; about 219 times 2025 earnings
- Agility Robotics
- SPAC merger with Churchill Capital Corp XI
- Signed 2026-06-24; close expected Q4 2026
- Trust $420.9 million plus PIPE $201.0 million
- Equity value $2.5 billion; proceeds depend on redemptions
- Persona AI
- Private offering, SEC Form D
- Filed 2026-09-14
- $65.36 million sold of $141.96 million
- 46 investors; a filing, not a priced round
Sources: SSE listing announcement (Unitree), Churchill Capital Corp XI filings with the SEC (Agility Robotics), SEC Form D (Persona AI). Not counted in the private-round totals above. Amounts as filed; no currency conversion.
As of Oct 15, 2026
MEKAWIRE analysis
In the month to October 14, private robotics money paid for the layers around the robot, chips, data, safety evidence and machines built for one job, while humanoid makers raised on exchanges; investors are now pricing the gap between a capable robot and a paid deployment rather than the robot itself.
The evidence is the 51.3% share for chips, the rise of data and safety software from 1.4% to 8.7%, and an October cohort in which every company sells a path to deployment. The investor overlap supports the second half: the same names that funded robot models are now funding the inputs those models need. The strongest counter-argument is the window. Thirty days is short, two chip rounds drive half the money and D-Robotics did not name a single investor. One humanoid round above $500 million before year end would undo the share figures, and our coverage of earlier months leans toward large deals.
Bull case: data and safety suppliers shorten pilots, humanoid makers that listed use public money to build fleets, and fourth-quarter rounds show the same tilt toward deployment layers. Bear case: the tilt is a lull between mega-rounds, Mecka's run-rate target slips, and public investors mark down humanoid makers once lock-ups and redemptions arrive. What would change our view is a private humanoid round above $300 million before December 31, which would show venture money returning to bodies. Three dated signals to watch: Unitree's third-quarter report due by October 31; the Churchill XI vote and redemption level in the fourth quarter; and MEKAWIRE's fourth-quarter refresh in early January 2027. The sharpest open question is whether Mecka can show the $300 million run rate it promised by year end, because that number would tell investors whether robot data is a business or a cost of doing one.
- 2026-10-15; figures disclosed, not estimated
- Sept 15 to Oct 14, 2026; 14 private rounds with disclosed US$ amounts
- 20.5% (General Intuition $220m)
- 51.3% (D-Robotics $400m, SiMa.ai $150m)
- Multiply Labs $75m, Mecka $60m, SafeWorld $12.2m, Minerva Humanoids about $10m
- $1,071.3 million
- 0.9% ($10m, Minerva Humanoids), down from 13.4%
- Unitree IPO RMB6.10bn gross; Agility SPAC $2.5bn equity value; Persona AI Form D $65.4m sold
- 3 of 4 October rounds share an investor with a robot-model or humanoid company
- 8.7%, up from 1.4% in Jan 1 to Sept 14
Why it matters
For founders raising in the fourth quarter, the window gives a disclosed benchmark of what investors are paying for. A company that can show it shortens deployment, with data, safety evidence or a validated process, raised in October on terms that humanoid makers did not get from venture investors in the same month. The 0.6 times ratio of Mecka's round to its claimed run rate also suggests that revenue, even unaudited, now decides the size of a round more than the scale of the vision.
For corporate buyers the shift lowers the cost of trying robots. Safety simulation, teleoperated first deployments and pay-per-output models such as Extend Robotics' let a plant test a robot without committing to a fleet. The companies that sell those bridges are the ones being funded.
Rival analysis
In robot data, Mecka competes with in-house capture programmes at the model labs and with motion-capture providers; Innodata opened a motion-capture laboratory for humanoid training in September. Its advantage is that it sells to many labs at once, and its risk is that the largest customers build their own. In safety, SafeWorld's closest analogue is the scenario-testing industry built for autonomous cars, and NVIDIA's Open Agent Safety Platform, launched in September, gives large robot makers an in-house alternative.
In biologics automation, Multiply Labs competes less with robot makers than with manual clean-room labour and contract manufacturers. Teradyne's stake gives it a hardware partner with global service, and AstraZeneca's gives it a reference customer type even though the release did not say AstraZeneca uses the system. Minerva's rivals are teleoperation-first companies such as Extend Robotics and Avatar Robotics, and the bomb-disposal robots already in police and military service.
Valuation context
Only two rounds in the window disclosed a valuation: SiMa.ai at $1.45 billion and General Intuition at $6.2 billion, the latter 2.7 times the $2.3 billion it disclosed 96 days earlier. None of the October four gave one. Against that, Unitree's listing priced the company at about RMB61 billion, 219 times 2025 earnings, and Agility's SPAC sets a $2.5 billion equity value on a company that reported $1.78 million of 2025 net sales in its S-4.
The gap between public humanoid valuations and private deployment-layer rounds is the clearest pricing signal in the data. Public investors pay for volume and listing scarcity; private investors in October paid for revenue claims and short paths to customers. If the two converge, it will be through humanoid makers buying the layers the October rounds are building.
Supply-chain implications
The chip money splits between China and the US. D-Robotics, which Horizon Robotics stopped consolidating in March, says its S600 part won more than 20 robot customers in six months, including UBTech, Fourier and Booster Robotics, which ties a large part of China's humanoid compute to one supplier. SiMa.ai plans a 1,000 TOPS chip for the first half of 2028 and sells IP and chiplets as well as finished parts.
The data layer is less exposed to trade policy but more exposed to labour: Mecka's capture fleets record people at work, which raises consent, privacy and wage questions in each market where it records. Minerva builds its first prototypes in Germany and said it intends to avoid single-source Chinese components, a design choice that raises cost at volume but suits defence and energy buyers.
Signals to watch
By October 31, 2026, Unitree's third-quarter report, the first public read on whether listed humanoid demand holds. In the fourth quarter, the Churchill XI shareholder vote and redemption level, which set how much cash Agility actually receives from the $420.9 million trust.
By December 31, 2026, Mecka's stated $300 million run-rate target and any named Fortune 50 customer for SafeWorld. In early January 2027, MEKAWIRE's fourth-quarter funding refresh, which will show whether the tilt toward chips, data and safety held or was a lull between mega-rounds.
Analyst view
Thesis: private money in the month to October 14 paid for deployment layers, not robot bodies. Confidence: medium. The share shift is large and the October cohort is consistent, but the window is short, two rounds drive half the total and one Chinese round has no named investors.
We would raise confidence if fourth-quarter rounds keep data, safety and application systems above 10% of disclosed money and if humanoid makers keep raising mainly through listings. We would lower it on a private humanoid round above $300 million or if Mecka's revenue claims are not borne out.
Questions you should be asking
For Mecka: what share of revenue comes from the largest customer, and how much of the run rate is data sales against robot integration? For Multiply Labs: which of the new investors are customers, and how many systems are in validated GMP use today?
For SafeWorld: has any regulator or insurer accepted its simulation results as evidence? For Minerva: who is the first paid pilot customer, and what is the operator-to-robot ratio in that pilot?