Softbank + 1X: the multi-billion dollar bet that could bring us the home robot

Mega-investor SoftBank is in talks to take a majority stake in 1X Technologies at a roughly $6 billion valuation, according to The Information, in an unconfirmed report Reuters has since echoed. Lots has yet to be decided: terms aren't final, and the size of the stake isn't known. 

Whether OpenAI — an early 1X investor that reportedly tried to buy the company outright last year — is selling isn't known either.

But here's what I know, because I've spent time with the robot and with the people building it: if this deal closes, Masayoshi Son will control one of the two or three most credible Western bets on a humanoid robot that actually lives in your house.

And 1X is about to need exactly the kind of money Son likes to shovel at potential category winners.

Son's whole career is this trade

Strip away WeWork and the Vision Fund's worst years, and Son's method has been consistent for three decades: identify a category he believes will be enormous, find the company he thinks wins it, and then hand that company so much capital that the question of whether it can outspend its rivals stops being a question.

In other words: anoint a winner.

He put $20 million into Alibaba in 2000 and turned it into one of the greatest venture returns in history. He bought Arm for $32 billion in 2016 when everyone thought he'd overpaid, and Arm is now the architecture layer under much of the AI compute buildout. He poured billions into Uber and DoorDash to buy market share in categories where scale was the moat.

To be fully forthright, he also lit $16 billion on fire in WeWork, and he sold SoftBank's $4 billion Nvidia stake years before the run-up that would have made it the best trade of his life.

No-one's perfect.

But he definitely likes robots. SoftBank bought Boston Dynamics from Alphabet in 2017 — then sold most of it to Hyundai in 2020, which might have . Last October he paid $5.375 billion for ABB's robotics business, one of the largest industrial arm makers on earth. He's backed Skild AI on the software side, and he's committed tens of billions to OpenAI. 

He has said, repeatedly and in public, that he expects a world with a billion humanoid robots. And so he's buying both brains and bodies. 

But what he doesn't yet have is a consumer humanoid — a robot for the house, not the plant floor.

That's 1X.

Why 1X, and why now

1X is Norwegian, with U.S. operations, and it made a decision early that separated it from almost everyone else in the sector: it went after the home first, instead of the factory.

That sounds like the easier problem. It is not. It's the harder one on every axis except one — data.

Chief Design Officer Dar Sleeper walked me through the logic when I interviewed him about NEO. Factory contracts demand near-perfect uptime; a home tolerates a robot that occasionally fumbles a sock. And a house generates messy, varied, unrepeatable training data that a repetitive industrial cell never will.

The hardware follows from that choice. NEO weighs a bit over 60 pounds, against 200-plus for many competitors, and it's tendon-driven rather than harmonic-drive: quieter, softer, safer to stand next to.

"If you have a harmonic-drive robot that's 150 to 200-plus pounds, it's very heavy," Sleeper told me. "You can't really be around people in a safe way."

The design language is deliberately un-robotic: soft materials, knit covering, Scandinavian restraint, hands that rest naturally at the sides instead of hovering in that uncanny half-clench most humanoids default to. "Neo definitely inherently takes on a personality," Sleeper said. He claims the adjustment period for people meeting it is short: "The time to normalize, believe it or not, is no more than a few minutes."

And it's not a home-only story, even though that's how it's been marketed. 

"Neo has always been for general labor," Sleeper said. "Neo's not just for the home. Neo's for everything." Last December, 1X signed a deal to put its home humanoid into factories and warehouses.

The hands

This is the part I'd underline for anyone evaluating whether $6 billion is cheap or expensive.

Humanoid hands are where the sector sometimes quietly fails. Plenty of companies ship impressive walking demos and then hand you a claw. 1X's hands are among the best I've seen in the industry, and Sleeper is precise about why in a way most spec sheets aren't.

"Twenty-two degrees of actuated freedom," he told me. "I think people don't really understand the differences here."

Then he explained the distinction: 

"22 degrees of actuated freedom means you can actually actuate every degree of freedom that is available." And: "There's 22 degrees of freedom versus 22 degrees of actuated freedom, and then there's also 22 degrees of freedom open loop or closed loop versus not."

Translation: a lot of published hand specs count joints that bend, not joints the robot can control. 1X counts the ones it drives.

That matters more than gait, more than battery life, more than almost anything else, because a home robot's entire value proposition is manipulation. Laundry. Dishes. Doors. Groceries. If the hands don't work, nothing else does.

He's also blunt about the economics underneath it: "We are able to achieve BOM costs that I don't think anybody else in the US game can achieve."

The shipping question

NEO is priced at $20,000. 

Preorders opened in October 2025 and 1X says the first year of production capacity sold out in five days. The company promised first deliveries within a year, and as of late August no customer shipments have been publicly confirmed. That deadline is coming up fast, and the gap is the single biggest thing skeptics point at.

But both Sleeper and CEO Bernt Børnich have personally told me they'll start shipping to customers this year.

"We're super excited to start shipping this year," Sleeper said. "I'm feeling good. I think the thing that nobody understands outside of being inside a humanoid company is just the deep complexity."

That complexity is the reason the SoftBank money matters. 

Getting from a working prototype to units in houses isn't a robotics problem at that stage ... it's supply chain, manufacturing yield, service logistics, replacement parts, customer support, even the design of the robot's clothing. 

Those are capital problems. They're the specific kind of capital problem Son has spent thirty years solving by simply removing the $$$ constraint.

What $6 billion buys

The $6 billion figure has been framed as a markdown from the $10 billion 1X was reportedly seeking last September. Be careful with that comparison: a majority control transaction and a minority venture round are different instruments, and control usually trades at a premium, not a discount. 

The real question isn't the valuation. 

It's whether the sector's window is opening or closing. Goldman Sachs projects annual humanoid shipments above 250,000 units by 2030 and a $38 billion market by 2035. Chinese manufacturers and Chinese state capital are flooding in, and they will likely win on price.

Western humanoids will have to win on something else: trust, safety, design, and hands that actually work in a kitchen.

That's a reasonable description of 1X's entire strategy. And if Son is right — if the home humanoid is the next platform rather than the next demo — then $6 billion for control of the leading Western candidate will look like the Alibaba trade, not the WeWork one.

I'd bet on Neo.