SoftBank's robotics spree points naturally toward construction autonomy, but Gravis Robotics should be treated as a strategic fit, not a confirmed takeover target.
Masayoshi Son is building a robotics portfolio with real machinery behind it. That part checks out. What doesn't check out is the published claim that SoftBank is weighing a deal for Gravis Robotics. I couldn't find a public report from Bloomberg, SoftBank, Gravis, or another reliable outlet confirming acquisition talks, so the story has to be narrower and more honest.
The cleaner version is still interesting. Gravis Robotics is exactly the kind of company that shows you where SoftBank's robotics map has a hole. The Zurich startup came out of ETH Zurich's Robotic Systems Lab in 2022 and builds retrofit systems for heavy earthmoving machines, including excavators and s. It doesn't ask contractors to buy a whole new fleet. It fits sensors, cameras, 3D mapping, real-time AI and hydraulic control onto machines that are already on site.
That is the useful detail. Construction sites are not factory floors. The ground changes, the work changes, and the machine you need today may not be the machine you need tomorrow. A robotic arm can live in a clean cell. An autonomous excavator has to work in mud, dust and bad geometry. Gravis is trying to solve that problem without asking the customer to rebuild the whole jobsite around the robot.
According to Gravis Robotics' November 2025 funding announcement, the company raised $23 million in fresh funding co-led by IQ Capital and Zacua Ventures, with Holcim, Pear VC, Sunna Ventures, Armada Investment and others participating. ETH Zurich also described the company as a D-MAVT spinout and said its systems combine 3D mapping and real-time AI to perceive the environment and directly control machine hydraulics. That gives the company a real base, not just a pitch deck.
SoftBank is buying the factory floor first #
SoftBank's confirmed robotics push is already large. On October 8, 2025, SoftBank said it had agreed to buy ABB's robotics business for $5.375 billion. ABB's robotics unit had about 7,000 employees and 2024 revenue of $2.279 billion, according to the transaction details SoftBank published. The deal is expected to close in mid-to-late 2026, subject to regulatory approvals in the European Union, China, the United States and elsewhere.
That is not small money. It gives SoftBank industrial robots, customers and distribution in one move.
Bloomberg also reported this week, as carried by Yonhap, that banks had finalized $1.75 billion of dollar and euro loans to back SoftBank's ABB Robotics acquisition. The banks named in that report were BNP Paribas, Goldman Sachs, HSBC and Mizuho. The financing detail matters because it shows the deal is still moving through the machinery, not sitting as a stale announcement from last autumn.
SoftBank has also been tied to Agile Robots, the Munich-based robotics company. Sifted reported in June 2026, citing Bloomberg, that Agile Robots was in talks to raise about $800 million and that SoftBank was discussing an investment of more than $300 million. Those talks were described as early, so you shouldn't treat them as closed. Still, the pattern is clear enough: SoftBank wants exposure to physical AI, not only software models sitting behind screens.
Gravis would fit that pattern. Fit is not proof. It is only fit.
Construction is the hard part #
Construction gives robotics companies a better excuse than most sectors: there are not enough workers, and the work itself is hard, dirty and dangerous. Associated Builders and Contractors said in January 2026 that the US construction industry needs to attract an estimated 349,000 net new workers in 2026 to meet demand. That is the right attribution. It is not a Bureau of Labor Statistics projection, and the earlier draft should not have called it one.
Gravis is selling directly into that shortage. Its own materials say the Gravis Rack can increase throughput by up to 30% with terrain-aware excavation, and the company says its systems are already being used across the UK, EU, US, Latin America and Asia. You should read company-reported productivity numbers with care, but they are still more useful than vague talk about automation changing construction. A contractor cares whether the trench gets dug faster and whether fewer people have to stand close to moving steel.
There are competitors too. Fortune reported this month that Amsterdam-based Monumental raised a $32 million Series B led by Khosla Ventures for bricklaying robots, and it noted Gravis as another construction robotics bet. Caterpillar, Komatsu and Volvo are also not going to ignore autonomous equipment. So the claim that nobody else has touched construction is too broad. The sharper point is that no one has yet made construction autonomy feel inevitable at scale.
SoftBank may decide it wants that piece. It may not. Frankly, the company already has enough robotics assets to keep management busy for years. ABB gives it industrial arms. Berkshire Grey and AutoStore put it near warehouse automation. Agile Robots, if that financing happens, would deepen the humanoid and industrial AI story. Gravis would add outdoor autonomy, but there is no verified public evidence that SoftBank is buying it.
That is where the article should land. Son's robotics empire is real, and so is Gravis' technology. The acquisition talk is not publishable unless someone has actually reported it.
Also read: MoonPay adds Discover Network cards to complete the US card trifecta for crypto buying, China's CXMT lists on July 27 in Asia's biggest IPO of 2026 and it's already moving markets, Amazon orders sellers to label AI-generated people in product images after New York law