AI Humanoid Robotics & Automation Fundraising (2026)

How Figure, 1X, Physical Intelligence, Skild, Symbotic, and humanoid, cobot, AMR, and VLA-model robotics startups raise capital in 2026 on GR00T.

Raising Capital for Humanoid, Industrial Robotics, and Warehouse Automation Startups

Robotics became the second-hottest AI category after frontier LLMs. Figure raised $675M Series B at $2.6B (Feb-2024) then $1.5B at $39.5B (Feb-2025). 1X Technologies $100M B (OpenAI-led). Physical Intelligence $400M at $2.4B (Nov-2024, Bezos + Thrive + Lux). Skild AI $300M at $1.5B (Jul-2024, SoftBank + Coatue + Bezos). Apptronik $350M (Feb-2025, B Capital + Capital Factory), Mercedes-Benz Apollo pilot live. Sanctuary AI, Agility Robotics (Digit at GXO Spanx facility), Unitree G1 China consumer, XPeng Iron, Tesla Optimus Gen-3, Boston Dynamics Atlas electric. On the industrial side, Symbotic $SYM (Walmart, Target, Albertsons), Locus Robotics, Nimble, Dexterity, Covariant-Amazon Aug-2024 talent-acquire, GreyOrange, Berkshire Grey-Softbank $375M take-private, RightHand Robotics-Palladyne acquired. VLA foundation models (Nvidia GR00T N1, Physical Intelligence π0/π0.5, Google RT-2/Gemini Robotics, Skild Brain) collapsed 5+ years of ROS integration work.

Why 2026 is different

Three unlocks converged: (1) VLA (Vision-Language-Action) foundation models — Physical Intelligence π0/π0.5, Nvidia GR00T N1 open weights (GTC Mar-2025), Google Gemini Robotics + Gemini Robotics-ER, Skild Brain, Figure Helix, 1X Redwood — collapsed the ROS/behavior-tree stack. (2) Hardware BOM crossed the $30-50k target for humanoid (Unitree G1 at $16k retail forced the whole industry down), and $50-150k for industrial cobot with 6DoF arms + dexterous hands (Sanctuary Phoenix 7, Apptronik Apollo, Figure 02). (3) Real deployments moved from PoC to production — Agility Digit at GXO (billable), Symbotic at Walmart 43 sites live + Target/Albertsons, Figure at BMW Spartanburg, Apptronik at Mercedes-Benz Sindelfingen, Sanctuary at Magna. US labor shortage (BLS quits rate + $18-32/hour warehouse wages) and CHIPS/IRA reshoring created a customer pull that didn't exist in 2019-2022. Franka Robotics insolvency (Feb-2024), Attabotics shutdown (Feb-2024), Berkshire Grey take-private, and Covariant → Amazon talent-acquire (Aug-2024) redistributed the losers into the winners' balance sheets.

Realistic capital stack

Seed: $10-40M for VLA foundation-model teams with lab pedigree (Physical Intelligence seed $70M, Skild seed $300M — outliers). Series A: $50-200M with a working robot + one commercial pilot. Series B/C: $200M-$1B with paying customers and hours-run telemetry (Figure B $675M, then C $1.5B; 1X B $100M; Apptronik B $350M; Physical Intelligence B $400M). Growth: Skild $300M at $1.5B (Jul-2024), Figure C $1.5B at $39.5B (Feb-2025 — highest robotics valuation ever). Public comps: Symbotic $SYM $20B+, AutoStore $AUTO Oslo listing, Ocado $OCDO LSE. Reference points: Agility Robotics $150M (Amazon, DCVC), Sanctuary AI ~$140M (Magna, Bell Canada), Boston Dynamics ($1.1B → Hyundai), Unitree ($100M+ from Meituan), XPeng Iron (parent XPeng $XPEV), Fourier Intelligence (China), UBTECH ($UBTECH HK IPO Dec-2023), Neura Robotics ~$120M.

Common failure modes

Building humanoid hardware without a foundation-model or teleoperation data flywheel (Franka Robotics insolvency Feb-2024, Halodi rumored distress). Chasing consumer humanoid without a labor-substitution ROI story (1X Neo Beta consumer bet is a live experiment). Warehouse AMR without SKU-agnostic pick-and-place capability (Attabotics shutdown Feb-2024). Selling perpetual-license industrial arms in a RaaS market. Ignoring EU Machinery Regulation 2023/1230 (effective Jan-2027) — replaces the 2006 Directive and adds AI-safety self-assessment. Under-provisioning training compute (Skild reportedly reserved 10k+ H100s pre-Series B). Assuming ROS 2 is enough — VLA models are eating ROS.

Frequently asked questions

Is humanoid a real market or a hype cycle?
Both. Real deployments exist (Agility at GXO billable, Figure at BMW, Apptronik at Mercedes, Sanctuary at Magna) but at pilot scale (10-100 units). Investors are underwriting the 2027-2030 curve to 100k+ units. Figure's $39.5B Feb-2025 mark is a bet on that curve, not on 2026 revenue. Expect valuation compression if hours-run telemetry disappoints in 2026-2027.
VLA foundation-model vs. full-stack humanoid — which is fundable?
Both, but different profiles. Foundation-model pure-plays (Physical Intelligence, Skild) fund like Anthropic — scaling-law bet, cross-embodiment thesis, licensing model. Full-stack humanoid (Figure, 1X, Apptronik, Agility) fund like Waymo — capex-heavy, deployment-milestone gated, RaaS revenue path. Middle tier (integration-only, no proprietary model, no hardware moat) is being squeezed.
Realistic exit?
Strategic acquisition by an OEM auto/appliance/robotics parent (Boston Dynamics → Hyundai $1.1B, Kiva → Amazon $775M, Kuka → Midea $5B, Covariant → Amazon talent, Berkshire Grey → Softbank $375M take-private, Franka → Agile Robots). IPO for category leaders (Symbotic $SYM, AutoStore $AUTO, Ocado $OCDO, UBTECH $UBTECH, RoboSense $RBTC, Hesai $HSAI). Foundation-model VLA pure-plays are AI-lab-style acquires or public offerings 2027+.

Related fundraising verticals (40)

Investor directory · Fundraising library · Articles A–Z · Company funding database