In 2025, 90% of all humanoid robots built and shipped worldwide came from China. Not 60%. Not 75%. Ninety percent. A new report from Bessemer Venture Partners puts a number on what the industry has been whispering about for months: the United States is losing the robotics race and it is not even close.

The investment gap is staggering. In the past five years, 745 software companies have raised more than $30 million each. The equivalent figure for robotics is 42. That is 18 times fewer funded companies in a sector where the underlying market is 30 times larger than global software spend.

"There will be 100,000x more robots on Earth in the next 10 to 20 years," says Jeremy Levine, a partner at Bessemer. Most analysts predict 50x growth in the robotics space over the decade. And yet the funding pipeline looks like a rounding error compared to software.

China Built the Factory. America Built the PowerPoint.

We have been tracking this divergence for months at The AI Post. China ships robots while the West ships press releases. AGIBOT put humanoids on a live electronics production line at 310 units per hour with 99.9% success rates. Honor's robot beat every human in Beijing's half-marathon by seven minutes. X Square Robot deployed cleaning teams in Shenzhen apartments. The Canton Fair showcased humanoids with prices down 30 to 40% year over year.

Meanwhile, Figure AI is worth $39 billion and just built 30,000 BMW X3s. That is genuinely impressive. But for context, Chinese startup AI2 Robotics is valued at $2.93 billion and is already rolling out robots at airports, semiconductor plants, and healthcare facilities. China gets more done at a fraction of the valuation.

CNBC reported today that Chinese startups are churning out humanoid robots faster than their US rivals despite far lower valuations. The country controls the supply chain, gets government backing through the 2026-2030 five-year plan targeting humanoid robots as frontier technology, and focuses on real customers at real prices.

Bessemer's Five Predictions

The report makes five core claims. The "ChatGPT moment" for robotics is coming but has not arrived yet. Scaled companies with capital will win because data is expensive and larger fleets learn faster. Fewer than 50 companies will capture the market, not hundreds. Full-stack players who vertically integrate hardware and software will take the most value, just as foundation model companies captured the LLM market. And defense robotics companies will have the first $50 billion IPOs, citing Anduril and Saronic.

That last prediction is worth sitting with. Bessemer is saying the first massive robotics exits will come from military applications, not consumer or industrial. That tracks with what we are seeing: nations have concluded that robotics fundamentally changes the nature of modern warfare. The Pentagon wants humanoid robots. So does China's PLA. The companies that build them will be worth more than most defense contractors.

The Data Problem Is Real

Not everyone is as bullish on the timeline. Lisa Yan, CEO of Argus Systems and a Waymo veteran, warned that closing the gap between 99% and 99.9% reliability "is a steep hill climb that takes longer than most people realize." Philipp Wu, co-founding a robotics company in stealth, said general-purpose robotics is still five-plus years out.

The industry collected 100,000 hours of robotic training data in 2024 and 1 million in 2025. The projected 20 million hours in 2026 still represents only 0.04% of what a general-purpose robot actually needs, according to UC Berkeley's Ken Goldberg. He calls it the "100,000-year problem."

But Bessemer's conclusion is that none of this matters for the investment thesis. The growth is coming whether the timeline is five years or fifteen. The question is whether American investors fund the companies that compete, or whether China builds the entire robotics supply chain the way it built the solar panel and EV battery industries. Right now, at 90% market share for humanoid shipments and 18x fewer funded US robotics companies, the answer is obvious.

America is barely trying.