The multiple worth writing down is 44. Contrary Research puts every dollar 1X has raised at $136.5 million [5], and a $6 billion price on the equity is about forty-four times that figure [1]. Venture funding and an acquisition price are different instruments, as the reporting itself notes, but the distance between them tells you what is being bought. Chinese companies held all five of the top shipment slots in the first half [3], so 1X, founded in Moss, Norway in May 2014 as Halodi Robotics and now run out of Palo Alto [4], is not on that table [9]. What SoftBank would be paying for is the thesis Contrary describes: the home as the training environment for general-purpose embodied intelligence [13].
Set the price against SoftBank's own recent robotics arithmetic. It agreed in October 2025 to buy ABB's robotics division for $5.375 billion, with closing expected during 2026 [10]. The reported 1X figure runs about $625 million above that, roughly 12 percent more [6], and the two together would come to some $11.4 billion of robotics acquisition [5]. The Vision Funds already hold more than 20 physical-AI positions carried at an unrealized fair value above $8 billion [11]; one $6 billion cheque would equal about three-quarters of that whole book [8].
The related-party geometry is worth reading slowly. The OpenAI Startup Fund led 1X's $23.5 million Series A2 in March 2023 alongside Tiger Global and Norwegian investors [6], and The Information reported in 2025 that OpenAI itself weighed buying 1X outright [7]. SoftBank has put more than $60 billion into OpenAI and related AI infrastructure [8], and its July 27 annual report carried the Vision Fund 2 stake in OpenAI at $79.6 billion as of the end of March [9]. On those numbers, $6 billion is 7.5 percent of the carrying value of the software position this hardware purchase is meant to extend [4].
Then the growth figure everyone will quote. Counterpoint said H1 2026 humanoid shipments passed 22,000, nearly 300 percent up year on year, with the full year possibly clearing 50,000 [2]. It did not publish an exact H1 2025 total, so the comparison base is inferred rather than disclosed [12]; near-quadrupling from 22,000 implies something around 5,500 units a year earlier [2]. Clearing 50,000 also requires the second half to beat the first by roughly 6,000 units [7]. And most of those units are not the market 1X is building for: AGIBOT and Unitree alone accounted for close to 75 percent of H1 volume [3], about 16,500 machines [3], while 1X aims Neo at domestic customers rather than businesses in factories [4].
So the volume evidence and the valuation thesis are measuring different markets. If the talks convert, $6 billion becomes the visible reference price for a home humanoid programme with no disclosed shipments at all [1], and the next founder raising against it will not have to explain the gap.
