Footage: Waymo, Anduril, Shield AI, Saronic, Figure, Unitree (official channels). Charts: behindrobotics.
Weapons won. Hands lost.
Physical AI raised $47.4B in H1 2026 across 521 deals (Crunchbase). Four rounds took 52% of it: Waymo $16B, Anduril $5B, Shield AI $2B and Saronic $1.75B. The 52% is our arithmetic on Crunchbase's totals; the article doesn't state it. Take Waymo out and the average deal was about $60M, flat on a year earlier.
PitchBook uses a different dataset and period (12 months to Q2 2026), so we keep it separate. There, defense & security was the #1 sector at $14.8B. Inside the robot stack, AI & autonomy platforms took $7.7B while end effectors (hands, grippers, tooling) got $444M: roughly 17:1.
"Lost" is relative. Hand funding still grew 426%; it just started from almost nothing.
The payoff case: about $100B of robot sales in 2025, and $2.5T by 2035 in the base case (range $0.5T to $8T). That $2.5T is an analyst average compiled by J.P. Morgan Private Bank from Morgan Stanley, BofA, Barclays, McKinsey and Bain, not J.P. Morgan's own forecast. Scenarios, not facts.
Sources
- Crunchbase News, 18 Aug 2026: Physical AI funding, H1 2026
- PitchBook, 1 Sep 2026: Q2 2026 Robotics & Physical AI Report (some sector figures via report summaries: fxbaogao, Global Economic)
- J.P. Morgan Private Bank: Robotics: the next big frontier in AI
- Our live datasets: funding scoreboard and database map