SoftBank Eyes 1X Control at a 40% Lower Valuation

The Information reported on August 26 that SoftBank is in talks to take a majority stake in humanoid robotics startup 1X Technologies, in a deal that would value the company at around $6 billion. Talks are ongoing and terms could change — the reporting says so itself, and nobody has signed anything.

But the story here is not “SoftBank buys another robotics company”. It is that number.

$6 billion, against the $10 billion it asked for

When 1X went out to raise last autumn, it sought a $10 billion valuation and a $1 billion round. What it actually got was less than half of that target.

A year later, SoftBank is discussing $6 billion — 40% below last year’s ask.

And this time the conversation is about a majority stake, meaning control, not a minority financial position. A startup moving from “I want to raise $1 billion at $10 billion” to “someone will take control at $6 billion” tells you more about the real temperature of humanoid robotics than any industry report.

In what is widely described as the hottest sector in technology, this is a down round.

Who 1X is, and what NEO is

1X is an OpenAI-backed company that built its profile on NEO, its humanoid robot. The company says it has received more than 10,000 orders.

That word deserves care in this sector. The figure comes from the company itself, and “orders” means wildly different things at different humanoid makers — it can be a fully paid, confirmed purchase, or a reservation list secured by a refundable deposit. Public information does not establish which category 1X’s 10,000 falls into, and there is no corresponding shipped-units figure to check it against.

This echoes something covered here before

This site reported in August on Tesla’s Optimus: Musk called it “the biggest product of all time”, while volume production stood, at that point, at zero.

Put the two together and the shape of humanoid robotics becomes clear: big claims, plenty of orders, very few units shipped. That is not a criticism of any individual company but a description of where the whole field currently sits. The technical demonstrations are genuinely convincing — walking, manipulating objects, interacting with people — but the distance from a demo to stable volume production, to units actually reaching customers and generating revenue, is still long.

Capital markets have evidently noticed the same thing. That is most likely why $10 billion became $6 billion.

“Majority stake” and “a funding round” are not the same thing

The word in this report that is easiest to skim past, and matters most, is majority stake.

Venture investment normally buys a minority position: cash in, preferred shares out, a board seat, but the founders keep control and the investor’s exit depends on a future IPO or sale. A majority stake means control changes hands: board composition, strategic direction, whether the company is sold and to whom, all become SoftBank’s decision.

For a hardware startup still burning cash without a product shipping at scale, founders entertaining a change of control usually signals one of two things: raising the next round independently has become hard enough that price alone will not fix it, or they have concluded that inside SoftBank — with its capital, manufacturing resources, and a sister business like ABB — they have a better chance of actually getting the product built.

These are not mutually exclusive, and both point at the same conclusion: the independent funding path is not looking good right now.

SoftBank’s other thread: the ABB robotics deal

Look only at 1X and this reads like bargain hunting. Widen the timeline and something else appears.

SoftBank had already agreed to acquire ABB’s robotics division for $5.4 billion, a deal expected to close during 2026. ABB’s robotics unit makes industrial arms — mature, revenue-generating machines that are actually running on factory lines today.

Masayoshi Son has called physical AI the company’s next frontier.

So SoftBank does not appear to be betting on one startup. It is buying both ends of the field at once: ABB representing commercialised, profitable but slow-growing industrial robotics; 1X representing the not-yet-commercialised, high-imagination humanoid end. The first supplies cash flow and manufacturing know-how, the second supplies the wager.

That combination is very much in character — using cash flow from mature assets to fund high-risk bets is SoftBank’s house style.

Why humanoids are especially hard

A lower valuation does not mean the field is failing, but it does reflect a reality coming into focus: the humanoid form is the hardest choice available.

Industrial arms have been commercially viable for decades because their problem is tightly bounded — bolted to the floor, repeating fixed motions, in a fully controlled environment, and if something goes wrong the line stops. A humanoid’s entire value proposition is the opposite: it must operate in environments designed for people, because that is what avoids rebuilding the factory or the house. Which also means coping with uneven floors, unfamiliar objects, people appearing at random, and every other contingency that cannot be enumerated.

Safety raises the bar further. A 100kg machine with two arms moving around near people faces a far heavier validation burden than an arm inside a cage. That cost and time do not fall proportionally as the AI models improve.

Hence the gap: the demos look close, and volume production is still far away. A demo needs to succeed once to become a video; production needs thousands of hours of uneventful operation in a customer’s home. The distance between those two is what the drop from $10 billion to $6 billion is pricing.

What is still unknown

Because the deal is not done, none of this is settled:

• Whether it closes at all. The reporting explicitly says talks are ongoing and terms may change. A meaningful share of stories like this never produce a transaction.
• What happens to OpenAI’s stake. 1X is OpenAI-backed, and there is no public information on how existing shareholders would be treated if SoftBank takes control. SoftBank is itself a major OpenAI shareholder, which makes the structure more tangled.
• NEO’s actual shipments. Without independently verifiable delivery numbers, there is no way to assess how much revenue 10,000 orders represents.
• Whether this valuation is the sector’s general level. One company’s down round does not prove the whole field is cooling; what matters is the terms other humanoid companies get over the coming months. 1X may simply be behind its own stated schedule.

How to read this

If your interest is the technology, this deal changes little in the near term — 1X keeps developing NEO, and SoftBank’s money most likely accelerates that rather than slowing it.

If your interest is the valuation cycle, the gap between $10 billion and $6 billion carries more information than any product launch. Humanoid demos keep getting better, and the price the market will pay for those demos has started coming down.

Both of those can be true at once. Technology curves and capital curves were never synchronised — the latter is far more sensitive to “how many more years”.

One last reminder about this story’s status: this is The Information citing sources, the deal is not complete, and neither SoftBank nor 1X has confirmed it. Until there is a formal announcement, $6 billion is a number under negotiation, not a price paid.