From Manus to Machines: What Humanoid Startups Should Learn Before 2026
Published:
Originally published on Substack.
In less than a year, Manus went from early-stage AI product to billion-dollar Meta acquisition.
I know because I was there — almost from day one.
Back in March, when the product launched, I gave it a task: research the future of humanoid robots. An AI agent studying the field I’d spent years working in.
Within days, the co-founder reached out directly. Invitation code. Eager for feedback. That kind of founder energy — the willingness to reach personally to early users — is something I’ve both practiced and recognized over years of building.
Less than ten months later, they’ve exited.
Clean. Strategic. Complete.
Not luck. A signal.
The Changing Economics of Exit
Old playbook: build, scale, survive, then maybe IPO or get acquired after proving durability.
New reality: small, focused teams exit quickly. Not because of revenue — because they become the obvious missing piece in someone else’s roadmap.
Manus didn’t win because it was the biggest. It won because it was the right puzzle piece at the right moment for Meta’s agent strategy.
The 2025 numbers tell one story:
Humanoid companies raised $1.71 billion across 16 rounds
Figure AI hit $39 billion valuation
SoftBank put $400 million into Agility Robotics
China: 168 companies funded, financing up 291% YoY
Yet Morgan Stanley warns of a coming “reset”. The gap between demos and deployments is widening.
Too much capital. Not enough clarity.
The Singapore Arbitrage
When I used Manus in March, the company was still operating from Beijing. By June 2025 — months before the Meta acquisition — they’d relocated headquarters to Singapore.
Not coincidental. Standard practice now for Chinese-origin AI startups with global ambition.
Singapore works as a translation layer: trusted by Western legal systems, familiar to Asian founders, neutral in geopolitical interpretation.
The lesson for humanoid startups is clear. If your hardware and software are designed for global deployment, your corporate structure must match. Not about tax. About trust, narrative, and acquisition readiness.
The Dexterity Trap
The humanoid race right now is obsessed with dexterity. Who can fold a shirt. Who can make a sandwich. Who can dance on stage.
Impressive feats. Also marketing. Not product-market fit.
What Manus understood — and what humanoid startups must learn — is that the real competition is cognitive, not mechanical.
The deciding factor is not whether a robot can do something impressive. It’s whether the right people can immediately understand what it does and why it matters.
Some startups are already applying this logic. In elder care, AroOne focuses not on cooking meals but on solving the
one child, two parents
problem — offering working professionals gentle presence, on-device privacy, and peace-of-mind notifications about aging parents. The narrative is simple enough for any investor to grasp:your parents stay independent, you stop worrying.
That simplicity — not technical sophistication — creates legibility.
What 2026 Demands
2025 was the “Year of Securitization”.
Deep Robotics is targeting IPO guidance by June 2026. Unitree aims for a$7 billion Shanghai STAR Market listing.
But most companies are still chasing demos, not deployments.
The priorities now are different. Pick one task and own it — if your robot can move totes safely for eight hours, that’s a milestone. Dexterity is a program, not a product. Build for integration, not independence; Tesla, Meta, and NVIDIA are looking for puzzle pieces, not competitors.
Restructure for global trust early. Singapore, Delaware, wherever — don’t wait until acquisition discussions. And master the narrative: agents, physical AI, embodied reasoning. These aren’t buzzwords. They’re conceptual handles that let outsiders map you into their roadmaps.
Above all, stay humble about hardware. Foundation models help robots see and reason better. But they still need great joints, power systems, structures. The teams that respect both AI and physics will move the industry forward.
The Shift
Traditional dream: scale.
Emerging reality: roadmap alignment.
When your company becomes the obvious fit in someone else’s strategy, acquisition stops being a transaction. It becomes alignment.
That’s what happened with Manus. The team didn’t chase an exit. They built something legible — and were discovered.
The most important skill for the next generation of robotics founders may not be engineering or fundraising. It may be translation — between cultures, markets, and ways of thinking.
Manus was discovered. The next one will be too.
But only if it speaks the language.
Further reading: Straits Times and Business Times Singapore on the Manus acquisition; 36Kr on China’s embodied intelligence funding surge; Morgan Stanley via Investing.com on the coming market reset.