EU clears SoftBank's $5.4 billion ABB Robotics deal, a price of 17 times EBITDA to give its robot AI a body
The European Commission cleared SoftBank Group's purchase of ABB's robotics division, valued at USD 5.375 billion, under its simplified procedure on 11 March 2026. The price equals 17.2 times the division's 2024 EBITDA, while the floor valuation of Skild AI, whose round SoftBank led in January, is 2.6 times as high. Approvals in China and the United States remain, and both companies expect closing in mid-to-late 2026.
SoftBank Group is paying about 17 times last year's EBITDA for ABB's robot arm business, a business whose earnings fell by almost a fifth in 2024. Two months before European regulators cleared the deal, the same investor led a funding round that valued Skild AI, a robot software start-up, at more than two and a half times that price.
The European Commission approved the USD 5.375 billion purchase on 11 March 2026 under its simplified merger procedure, 26 days after the parties notified it on 13 February, according to the decision in case M.12248. Brussels was one of three approvals SoftBank listed when it signed the agreement, alongside China and the United States. What SoftBank is buying, and why the price looks modest next to the software valuations it is underwriting at the same time, says a good deal about where investors now think the money in robotics sits.
The price is set by a business that shrank in 2024
SoftBank said on 8 October 2025 that ABB would carve its robotics division into a new holding company and that a SoftBank subsidiary would buy all of its shares for USD 5.375 billion, about JPY 818.7 billion, subject to the usual adjustments for working capital and net debt at closing. ABB said it expects about USD 5.3 billion in net cash proceeds after transaction costs, a non-operational pre-tax book gain of about USD 2.4 billion, separation costs of about USD 200 million and transaction-related cash tax outflows of USD 400 million to 500 million.
The pro forma figures SoftBank published show the trajectory. Revenue fell from USD 2,452 million in 2023 to USD 2,279 million in 2024 and EBITDA from USD 385 million to USD 313 million. By ROBOTNESS arithmetic that is a 7.1% fall in sales and an 18.7% fall in EBITDA, taking the margin from 15.7% to 13.7%. ABB, using its own measure, put the division's 2024 operational EBITA margin at 12.1% on revenue of USD 2.3 billion, 7% of group sales, with about 7,000 employees and headquarters in Zurich. The enterprise value works out at 2.36 times 2024 revenue and 17.2 times 2024 EBITDA.
- Revenue (pro forma, unaudited)
- 2452
- 2279
- EBITDA (pro forma, unaudited)
- 385
- 313
- EBITDA margin (%)
- 15.7
- 13.7
- Enterprise value / revenue (x)
- No data
- 2.36
- Enterprise value / EBITDA (x)
- No data
- 17.2
Revenue and EBITDA from SoftBank Group's 8 October 2025 announcement. ROBOTNESS arithmetic: margin = EBITDA / revenue; multiples = USD 5,375m enterprise value / 2024 figure. ABB's own 2024 operational EBITA margin for the division was 12.1%, a different measure.
As of Oct 1, 2026
Those multiples describe a mature, cyclical machinery business. The regulatory record explains why the deal moved through Brussels without friction.
What Brussels cleared, and what is still open
The Commission's decision, C(2026) 1804, was adopted under Article 6(1)(b) of the Merger Regulation using the simplified treatment, and the public version describes the parties in a sentence each. SoftBank appears as a listed investment holding company with technology investments and subsidiaries, and ABB Robotics Holdco 1 Ltd. as the vehicle that designs, makes, sells and services robots. The published text contains no market definitions and no market shares. From signing on 8 October 2025 to the Brussels decision took 154 days.
China and the United States were still outstanding at that point, according to the list of approvals SoftBank gave at signing. ABB, reporting full-year results on 29 January 2026, repeated that the sale remained subject to regulatory approvals and other customary conditions and was expected to close in mid-to-late 2026. The same release booked USD 123 million of stranded costs in 2025 linked to the pending divestment. ABB's group orders rose 17% to USD 36.8 billion in 2025 and revenues 9% to USD 33.2 billion, at an operational EBITA margin of 19.0%, and the company guided to comparable revenue growth of 6% to 9% in 2026. The division's 12.1% operational EBITA margin in 2024 sat well below the 19.0% the group reported for 2025, which is the context for ABB's choice.
Why ABB took a cheque instead of a listing
ABB had prepared to spin the division off as a separately listed company and dropped that plan when SoftBank made its offer. The company said robotics had limited business and technology synergies with the rest of ABB and different demand and market characteristics. Chairman Peter Voser said SoftBank's offer reflected the division's long-term strengths and would create immediate value, and chief executive Morten Wierod pointed to SoftBank's capabilities in AI. ABB keeps its Machine Automation division and said in October it would fold it into its Process Automation business area.
For ABB shareholders the case is arithmetic. For SoftBank the case rests on what it intends to put inside the robots.
SoftBank wants a body for its robot software
Masayoshi Son, SoftBank's chairman and chief executive, framed the deal as part of what he calls Physical AI, saying the group would unite ABB's technology and people under a vision of fusing artificial super intelligence and robotics. SoftBank described AI chips, AI robots, AI data centres and energy as its four strategic pillars. On 14 January 2026 it led Skild AI's USD 1.4 billion Series C, which Skild said valued it at more than USD 14 billion, with NVIDIA's NVentures, Macquarie Capital and Bezos Expeditions joining.
The mechanism is straightforward. A general robot model like Skild's improves with data from real deployments and earns money only when it runs on machines that customers already trust. ABB Robotics brings an installed base of industrial arms, a service organisation and production capacity. SoftBank has been here before in a different direction. In June 2021 it sold control of Boston Dynamics to Hyundai Motor Group in a deal that valued the US company at USD 1.1 billion, keeping 20% through an affiliate. The ABB price is 4.9 times that valuation.
A rival set that is mostly industrial
ABB Robotics competes in industrial arms with FANUC, Yaskawa Electric, KUKA and Kawasaki Heavy Industries, and in collaborative robots with Universal Robots, which belongs to the US test-equipment maker Teradyne. FANUC, Yaskawa and Kawasaki are listed Japanese machinery makers. Once the deal closes, ABB Robotics will be the only top-tier arm maker controlled by a technology investor, and that changes what its sales force can promise customers about software.
ROBOTNESS analysis
SoftBank is buying ABB Robotics at an industrial multiple to give its robot software bets a body, and the gap between the two prices shows how much more investors now pay for robot intelligence than for robot hardware.
The evidence sits in one quarter of deal flow. SoftBank pays 2.36 times revenue for a profitable maker with roughly 7,000 staff, while the floor valuation of the software company it backed in January is 2.6 times the ABB enterprise value. Mobileye paid USD 900 million in January for Mentee Robotics, a humanoid developer that targets customer pilots in 2026 and series production in 2028, a sixth of the ABB price for a company without series revenue.
- ABB Robotics
- Sale of ABB's robotics division to SoftBank Group (EU clearance 2026-03-11)
- 2025-10-08
- 5375
- 5,375
- Enterprise value
- Skild AI
- Series C led by SoftBank
- 2026-01-14
- 1400
- 14,000
- Post-money, stated as 'over' this floor
- Mentee Robotics
- Acquisition by Mobileye, cash and stock
- 2026-01-05
- 900
- 900
- Purchase price
- Apptronik
- Series A-X extension
- 2026-02-11
- 520
- No data
- Valuation not disclosed
- Mind Robotics
- Series A, co-led by Accel and Andreessen Horowitz
- 2026-03-11
- 500
- No data
- Valuation not disclosed
- LimX Dynamics
- Series B
- 2026-02-02
- 200
- No data
- Valuation not disclosed
Figures as disclosed by the companies on their own pages, not estimated. Skild AI's valuation is a floor ('over $14 billion'). Ratio cited in text: Skild AI floor valuation / ABB Robotics enterprise value = 14,000 / 5,375 = 2.6x.
As of Oct 1, 2026
The strongest counter-argument is that the ABB price reflects the division's own decline rather than any discount on hardware. Sales and EBITDA both fell in 2024 and industrial robot demand is tied to automotive and electronics capital spending. On that reading SoftBank paid a fair cyclical price and the Skild comparison says more about venture exuberance than about a structural shift.
Bull case. Closing in the second half of 2026 gives SoftBank a global arm maker on which to deploy foundation models, and ABB's service relationships turn model improvements into recurring software revenue. A margin that has fallen to 13.7% has room to recover if software lifts pricing.
Bear case. Integration of a manufacturer is operational work that an investment holding company has rarely done at this scale, and SoftBank's earlier robotics holdings changed hands. If Chinese or US approval brings conditions, or if customers hesitate over ownership by an AI investor, the business could keep shrinking while the software story waits.
Signals to watch:
- ABB's first-quarter 2026 results in April, for stranded costs and any update on approvals.
- A decision by China's market regulator, which SoftBank named as a required approval, before the targeted mid-to-late 2026 closing.
- The first joint product announcement after closing that puts a SoftBank-backed model on an ABB controller.
The open question is whether SoftBank keeps ABB Robotics as an arm maker that buys AI or turns it into the hardware channel for the models it has funded. The answer will show in the first product roadmap after closing.
- SoftBank Group buys ABB's robotics division
- Cleared 11 March 2026, case M.12248, simplified procedure
- 13 February 2026
- 17.2x (ROBOTNESS arithmetic)
- USD 5.375 billion (about JPY 818.7 billion)
- Mid-to-late 2026
- About USD 2.4 billion pre-tax
- USD 313 million (pro forma)
- USD 2,279 million (pro forma)
- EU, China, United States
Why it matters
The EU clearance moves the largest robotics transaction of the past year closer to completion and confirms that Brussels sees no competition issue in an AI investor owning a top-tier industrial arm maker. That precedent matters for any later combination of foundation-model owners and hardware makers in Europe, because the public decision records no market definition or share analysis at all.
For the industry it creates a new type of owner. FANUC, Yaskawa and Kawasaki answer to public shareholders who judge them on machinery margins. A SoftBank-owned ABB Robotics can be judged on how much software it carries, which could change pricing, service contracts and partner choices across the arm market.
The deal also fixes a public price for a mature robot business at 2.36 times revenue and 17.2 times EBITDA, a reference point that will be used when other industrial robot units are valued.
Rival analysis
FANUC and Yaskawa are the closest comparables in industrial arms, both independent and listed in Tokyo, and neither has an AI investor as its owner. Kawasaki Heavy Industries runs robots inside a diversified group. Universal Robots competes in collaborative robots under Teradyne, and KUKA is the main European rival. None of them has an owner that also funds a robot foundation model.
The near-term threat to rivals is limited. SoftBank still needs Chinese and US approval and has not disclosed a technology roadmap. The medium-term risk is that ABB Robotics bundles models from SoftBank portfolio companies at a price rivals must pay partners for, which would pressure their software margins rather than their hardware share.
Korean entrants such as Doosan Robotics and Rainbow Robotics, and Chinese collaborative-robot makers, gain a different argument with customers who prefer an industrial owner. Expect that point to appear in sales pitches.
Valuation context
At USD 5.375 billion the enterprise value equals 2.36 times 2024 pro forma revenue of USD 2,279 million and 17.2 times EBITDA of USD 313 million. Measured against 2023 EBITDA of USD 385 million the multiple would be 14.0 times, so the price looks fuller on the depressed year. ABB's expected book gain of about USD 2.4 billion indicates the carrying value it held for the business.
Set against venture marks in the same quarter, the hardware is cheap. Skild AI's valuation floor of more than USD 14 billion is 2.6 times the ABB price. Mind Robotics raised USD 500 million in its Series A in March and Apptronik USD 520 million in February, single rounds that each equal about a tenth of the ABB price. The Boston Dynamics valuation of USD 1.1 billion in 2021 is the other reference SoftBank knows directly.
ROBOTNESS does not estimate a value for ABB Robotics under SoftBank. All figures above are disclosed by the companies or computed from disclosed figures.
Supply-chain implications
An industrial arm depends on precision reducers, servo motors and drives, controllers and safety electronics. A change of owner does not alter those dependencies on day one, and neither company has disclosed any change to sourcing or manufacturing sites. Procurement is therefore a question for after closing rather than a condition of it.
The approvals SoftBank named, EU, China and the US, map onto where the business sells and builds. China is both a major market for industrial robots and the home of fast-growing domestic arm makers, so its review is the one most likely to shape timing.
The software layer is where SoftBank adds exposure. Running foundation models on robots requires on-board compute, and SoftBank names AI chips as one of its four pillars, which makes the compute supply chain a strategic question for ABB Robotics in a way it was not inside ABB.
Signals to watch
ABB's results for the first quarter of 2026, due in April, should update stranded costs and approval status. A Chinese clearance, with or without conditions, is the main gating item before closing in mid-to-late 2026.
After closing, watch for leadership changes at ABB Robotics, the first SoftBank portfolio model offered on ABB hardware, and any announcement on the Zurich headquarters or the workforce of about 7,000.
Rivals' responses are a third signal. A tighter AI alliance between FANUC or Yaskawa and a model developer within six months of closing would show the market treating SoftBank's move as a threat.
Analyst view
Thesis: SoftBank is buying a mature arm maker as the hardware channel for its robot software bets, at a price that reflects both the division's 2024 decline and a market that now pays far more for robot intelligence than for robot hardware.
Confidence: medium. The deal terms and the EU decision are fully disclosed and the valuation gap is a matter of record. What remains unknown is SoftBank's integration plan; neither company has disclosed how SoftBank models will reach ABB products, and the Chinese and US reviews were still pending when Brussels ruled.
What would change our view: a closing announcement that keeps ABB Robotics at arm's length from SoftBank's AI portfolio, or conditions from Chinese regulators that limit technology transfer, would make this an ordinary financial acquisition rather than a strategic bet.
Questions you should be asking
For SoftBank: Which portfolio models, if any, will run on ABB controllers in the first year after closing, and on what commercial terms? Will ABB Robotics remain headquartered in Zurich with its current management?
For ABB: What drove the 18.7% fall in the division's EBITDA in 2024, and how much of the USD 123 million in stranded costs will remain after closing?
For both: What, if any, remedies have Chinese or US authorities asked for, and does the timetable still point to mid-to-late 2026?