←Reports

Humanoid Business Impact

Read full report →Request full PDF
Report·Humanoid·Mid-2026
Humanoid
Business Impact

The business case for humanoids in mid-2026: where the capital is, what underwrites it, and whether the bet pays off.

3 Takeaways
01
Capital has run far ahead of revenue: an 8.2% revenue-to-investment ratio. The market is being built ahead of the revenue needed to justify it.
02
The demand originates from structural labor shortages: 1.9 million unfilled US manufacturing jobs by 2033, an 11 million worker shortfall in Japan by 2040.
03
The real risk is timing, not whether the technology works. The payoff at scale is not expected until around 2032, so if investors pull back before revenue catches up, a humanoid winter will arrive.
The core tension
01 / Capital vs revenue
  • Humanoid venture funding hit ~$6.1 billion in 2025 against ~$500 million in revenue.
  • An 8.2 percent revenue-to-investment ratio: the market is built ahead of its demand.
Capital raised (2025)~$6.1B
Global revenue (2025)~$0.5B
2025 revenue is roughly 8.2% of capital raised
Demand foundation
02 / Labor scarcity
  • The bet rests on a structural, primary-sourced labor shortage.
  • Structural, not cyclical, which pulls toward permanent automation.
1.9M
unfilled US mfg jobs by 2033
3.8M
US workers needed this decade
11M
Japan worker shortfall by 2040
Why humanoid
03 / General vs special
General beats special
  • The unfilled roles are varied work in human-built spaces, not what a fixed arm does.
  • One machine that does ten jobs beats ten machines that each do one, if the AI generalizes.
Special-purpose
Fixed robotic arm
DoesOne task, in one place
EnvironmentA purpose-built workcell
Re-taskingRe-engineer the cell per task
Wins whenA single task has high volume
General-purpose
Humanoid
DoesMany jobs, switches between them
EnvironmentHuman-built spaces
Re-taskingSame machine, new task
Wins whenNo single task has enough volume
Two markets
04 / Capability vs volume
Capability-first vs volume-first
  • The West optimizes capability; China optimizes volume and price.
  • The two may converge as Western costs fall and Chinese capability rises.
VOLUMECAPABILITYconvergingWesterncapability-firstChinesevolume-first
The capital race
05 / Power-law concentration
  • Funding is power-law concentrated in a few Western names.
  • The sector has reached public markets (Agility SPAC, Unitree IPO).
Western private capital raised
Figure AI$1.845B
Apptronik$935M
Agility Robotics$640M
Power-law concentrated. Three names hold most of the private capital.
Do the economics work
06 / Price and labor offset
  • Hardware price spans an order of magnitude and is falling.
  • The robot crosses the cost line today only in high-wage, structured settings.
Hardware price, per unit
Unitree G1~$16K
Apptronik Apollo (target)<$50K
Agility Digit~$250K
Pricing already spans an order of magnitude and is falling
Cost per hour, labor offset
Human (fully loaded)~$30/hr
Digit today$10-12/hr
Digit at scale (target)$2-3/hr
Crosses the line today only in high-wage, high-turnover, structured settings
Market size
07 / The forecast spread
  • 2035 forecasts span more than a hundredfold, from ~$15B to the trillions.
  • Small numbers bet on factories, large numbers on living rooms.
100x+ SPREAD$15BInteract$38BGoldman~$5TMorgan Stanley$10B$100B$1T$10TFACTORIESLIVING ROOMS
Log scale. Small numbers bet on factories, large numbers on living rooms.
The bubble question
08 / What to watch
What to watch
  • ↓A down round or failed raise by a top-tier Western player would mark the tightening that triggers a shakeout.
  • ↑Recurring revenue disclosures converting backlog into income, such as Agility’s contracted Digit orders becoming reported revenue, would de-risk the thesis in the other direction.
  • •The first verified headcount-reduction figure from a flagship Western deployment would be a hard number showing that a deployment let a company employ fewer people. As of mid-2026, no such figure exists.
Protocol
Research toward a safer autonomous world
protocolz.org
01 / 11