$30 Billion in Funding in 3 Months, But I Think Humanoid Robot "Mass Production" Might Flop

Last week when I saw Xinghai Chart raised another 2 billion, my first reaction wasn’t “awesome” but rather —

“That’s burning cash way too fast.”

Honestly, the embodied intelligence track at the start of 2026 can no longer be described as “hot” — it’s “scorching.” Xinghai Chart with two large funding rounds within two months, valuation directly hitting 20 billion; UBTECH planning to spend $18 million recruiting AI scientists; Zhi Square, Qianxun Intelligence, Independent Variable Robot all entering the “10-billion unicorn” club. Roughly calculated, Q1 total funding for the entire track is nearly 300 billion.

What’s this concept?

Roughly equivalent to a medium-sized city’s annual GDP.

But the question is — the money came in, where’s the product?

I’ve done algorithms and seen plenty of “PPT fundraising” routines. A friend at a robotics company told me their goal this year is “mass produce 100 units.” I asked who they’re planning to sell these 100 to; he paused and said “mainly for B-end client pilots.”

See, “pilots” again.

This reminds me of new energy vehicles a few years ago. Back in 2019, NIO, XPeng were posting daily battle reports saying deliveries broke 10,000, but how many did you see on the streets? How did they rise later? Charging station density increased, supply chain matured, prices came down.

Humanoid robots’ current situation is somewhat like new energy in 2019 — technically not that they can’t run, but far from “ordinary people can afford and use.”

What’s the core problem?

First, data. You want a robot to go to a coffee machine and make you an Americano — how much data is needed to train this action? Conservatively estimated, millions of demonstrations. The industry-standard approach now is — buy real human motion capture data. But the cost… those who’ve done it know.

Second, supply chain. Harmonic reducers, torque sensors, hollow cup motors… these precision components’ domestication rate is constantly improving, but high-end products are still dominated by Japan and Europe. A humanoid robot’s BOM cost is estimated at 300-500k RMB. Do you think ordinary consumers will spend 500k on a “robot that can help get deliveries but might fall over”?

Third, and most critical — real-world scenario validation. Factory assembly lines could potentially use them, but what’s a worker’s monthly salary? Factor in robot depreciation and maintenance costs, ROI might still be negative.

So when I see “2026 is year one of humanoid robot mass production,” my feelings are mixed.

Not saying it’s impossible, but “year one” is used too early. The current excitement looks more like capital grabbing entry points — occupy the track first, profitability comes later.

That said, history always rhymes.

In 2019, new energy was criticized as “subsidy fraud.” In 2024, new energy penetration exceeded 50%. Maybe five years from now, looking back at today’s embodied intelligence investment frenzy, we’ll find those who bought at the bottom really made a fortune.

The question is — do you dare to bet?

I’m still watching for now. Not that I don’t believe in this direction, but I feel entering now is a bit like buying LeEco stock in 2015. Is it a good company? Maybe. Is the timing right? That’s debatable.

By the way, if your company is considering introducing robotics solutions, I suggest calculating ROI first — not listening to suppliers’ promises, but pulling up a three-year cost comparison table in Excel yourself. After calculating, you might not want to buy anymore.