Robotics and physical AI startups raised $8.73 billion across 32 disclosed equity rounds in the 12 months through July 2026, with follow-on rounds dominating deal-making: 92% of deals and 94.2% of capital went to existing companies rather than new entrants. The category is consolidating around proven winners—Figure AI ($48B), Skild AI ($14B), Physical Intelligence, Generalist ($2B), and Apptronik ($935M in Series A)—while new-company formation has become a specialist play around bottleneck layers like perception, simulation, and training data.
The market split into two thesis camps: humanoid robot companies (17.1% of funding) and foundation-model labs (robot brains). Robot foundation models commanded higher valuations relative to deployment proof: Skild AI tripled to $14 billion in seven months, Physical Intelligence and Generalist both priced north of $5 billion despite limited public shipping data. By contrast, warehouse automation—the mature physical AI segment—pulled in $3.6+ billion across eight companies but commanded lower multiples, signaling investor preference for architectural leverage over applied verticals.
For architects: capital has frontloaded toward platform contenders. The barrier to entry for a credible robotics startup now requires hardware depth, data strategy, safety credibility, manufacturing access, deployment partners, and frontier AI talent—a combination that excludes bootstrap plays. Strategic investors (Amazon, Aramco, SoftBank, NVIDIA, Toyota Ventures) are co-investing alongside VCs, signaling that physical AI is now table-stakes infrastructure, not moonshot hardware. Expect the long tail of single-vertical robotics startups to face compressed returns.