China reportedly blocks SoftBank’s $5.375 billion ABB Robotics deal

By Rebecca Stone

Chinese authorities have reportedly blocked SoftBank’s proposed $5.375 billion acquisition of ABB’s robotics business. The Financial Times reported the decision on the 5th, local time, citing multiple sources, and said regulators had notified SoftBank that the deal would not be approved.

The reported block interrupts a transaction announced last October and previously expected to close in mid-to-late 2026, subject to regulatory approvals in several countries. Valued at $5.375 billion, or approximately 7.4 trillion won, the purchase would place substantial capital behind SoftBank’s robotics strategy, although the source provides no breakdown of its funding.

Which regulator reportedly stopped the deal?

The notification reportedly came from the National Development and Reform Commission (NDRC), the body that oversees supply-chain security and stability reviews. SoftBank had already passed China’s antitrust review in February, but that clearance did not resolve the separate national security and supply-chain assessment.

Passing one review does not substitute for another in this transaction. Under the reported sequence, February addressed the competition gate, while the later review tested whether the ownership transfer and associated technology flows created a strategic risk.

That distinction changes how R&D and finance teams should read the October timetable. The original close date assumed approval from multiple jurisdictions, so Chinese clearance was necessary but not sufficient. Technology, intellectual property (IP) and operating data also remained central to the reported review, not merely post-closing integration choices.

Why did robot data become a security issue?

Chinese authorities were reportedly concerned that robotic technologies and related data developed and accumulated by ABB in China could be transferred abroad. Their reported focus included whether robot data from within China could train models for overseas artificial intelligence (AI) companies such as OpenAI. SoftBank’s status as a major investor in OpenAI gave the issue a direct strategic dimension.

The concern reaches beyond the transfer of industrial assets to the destination of information produced by robotic operations. A change in shareholder control does not by itself determine where data generated in China can be reused.

That distinction places the review beyond a change-of-control assessment. It also defines what the evidence can and cannot establish: the source identifies no particular dataset, model architecture or training run, so it supports concern about a possible pathway, not a finding that a specific overseas transfer had occurred.

What remedies were reportedly considered?

During the NDRC review, officials reportedly proposed conditions that would:

• Sell parts of ABB’s robotics business to Chinese owners.
• Keep certain intellectual property rights within China.

The source does not specify whether these were alternatives, mandatory remedies or terms accepted by either party. Their reported appearance nevertheless places ownership and the location of protected IP inside the approval test, rather than treating them as purely post-closing integration matters.

What is confirmed, and what remains reported?

Neither Chinese authorities nor the parties involved had issued official announcements confirming the rejection or the review details. The Financial Times account cited multiple sources, but the available report does not state how many people were interviewed or identify them. For decision-makers, the block must therefore remain a reported regulatory outcome, not a confirmed agency record.

The absence of official notice does not make the review immaterial. The February antitrust clearance and the later national security review tested different questions, and passing the first did not remove the second. A portfolio plan that treats a completed competition approval as evidence of broad cross-border clearance would misread the sequence.

What changes for R&D and portfolio plans?

For SoftBank, the reported veto is likely to create setbacks for its “Physical AI” business vision, meaning the strategy that links physical robotic systems with AI investment. The $5.375 billion acquisition was the concrete portfolio action at issue.

Without Chinese approval, planned spending, governance and integration schedules tied to a mid-to-late 2026 close must carry a higher probability of delay or redesign. The source discloses no revised timetable, termination charge, alternative financing package or negotiated carve-out, so none of those can be treated as settled.

For R&D managers, the immediate diligence task is to map data rights and cross-border security gates alongside conventional valuation and competition analysis. The reported remedies imply that any revised case would have had to address more than a new closing date: ownership, technical control and IP would each require a defensible allocation consistent with China’s security concerns.

SoftBank’s next decision point is whether to pursue the reported Chinese-ownership or domestic-IP alternatives, because without renewed approval the acquisition cannot close on the announced terms in mid-to-late 2026.