Analysis
Robotics startups have raised roughly $23 billion so far in 2026, closing in on the full-year 2025 total, with humanoid-specific funding alone crossing $8.6 billion -- 1.8 times last year's full-year figure -- according to Crunchbase News' sector data and Briefs.co's tracking.
I think the honest read on this is that humanoid robotics is being priced like a software category with hardware attached, and that math only works if a very specific set of assumptions holds.
Look at the specific checks:
“Look at the specific checks: - Figure AI -- raised roughly $2.34 billion cumulatively at a $39 billion valuation.”
- Figure AI -- raised roughly $2.34 billion cumulatively at a $39 billion valuation.
- Neura Robotics -- closed a financing worth up to $1.4 billion, with Nvidia, Amazon, Bosch and Schaeffler at the table.
- Apptronik -- extended its Series A to more than $935 million total.
These are venture-scale multiples on companies that, as of today, sell almost no commercial units at positive gross margin. The bet embedded in every one of these checks is that humanoid hardware follows a cost curve like batteries or solar panels -- falling fast enough, fast enough, that a company burning hundreds of millions today is buying market position for a business that becomes genuinely profitable at scale within a few years, not a decade.
That bet has worked before. It worked for electric vehicles, eventually, after a decade of capital destruction that wiped out most of the first wave of EV startups before Tesla and Chinese manufacturers proved the curve. It has not yet worked for any humanoid robot company, because none has shipped units at the volume needed to prove the curve exists in this category specifically. Battery cost curves benefited from a decade of consumer electronics demand pulling manufacturing scale forward before EVs ever needed it. Humanoid robots have no equivalent adjacent high-volume market subsidizing their component costs -- actuators, dexterous hands and vision systems are being built mostly FOR humanoid robots, which means the cost curve has to be earned inside this category alone.
Room for disagreement: the strongest counter is that humanoid robots don't need their own volume to benefit from a cost curve -- industrial automation, warehouse robotics and EV manufacturing are already driving down the cost of the exact components humanoids use, from servo motors to lidar to battery cells. If that's true, humanoid-specific volume matters less than I'm giving it credit for, and the current capital is buying a real, near-term cost advantage rather than a hopeful one. Nvidia, Amazon and Bosch backing Neura Robotics specifically, rather than only pure financial investors, is real evidence for this case -- strategics with manufacturing expertise don't usually write checks on a decade-out thesis without believing the near-term unit economics are closer than they look from outside.
What I'd actually diligence before writing a check in this category: unit cost trajectory per generation, disclosed with real numbers rather than a roadmap slide, and whether a company's manufacturing partner is the same one supplying components to a higher-volume adjacent category. The company that can show its actuator costs fell because of someone else's manufacturing scale, not its own promised volume, is the one worth the software-style multiple.
