China has blocked Swiss industrial group ABB from selling its robotics division to Japan’s SoftBank Group, Bloomberg reported on Nov. 5.

ABB offered more detail in an emailed statement: China’s Office of the Working Mechanism for the Security Review of Foreign Investment, under the National Development and Reform Commission, had blocked the $5.4 billion deal. SoftBank founder Masayoshi Son had viewed the acquisition as a key part of his effort to expand the group’s influence in AI infrastructure, and the decision has dealt that plan a setback.

SoftBank shares fell about 7% in Tokyo on Thursday following the news. A SoftBank spokesperson declined to comment.

A Deal Between Switzerland and Japan

Start by looking at the two sides of the deal.

The seller, ABB, is a Swiss industrial group, while the buyer, SoftBank, is a Japanese investment group. The target is ABB’s robotics division, which employs more than 7,000 people, supplies robotic arms and industrial equipment to automakers including BMW, and has extensive operations in China. The $5.4 billion price tag was not for a technology license, but for an entire business complete with employees, production lines and customers.

Ordinarily, the sale of a Swiss company’s business unit to a Japanese group would appear to have little direct connection to China. Yet China is precisely where the deal ran aground.

The Target, Not the Buyer or Seller, Was the Sticking Point

In cross-border M&A, the approvals a deal requires are determined not by where the buyer and seller are registered, but by the jurisdictions where the target has a physical presence.

This is easy to overlook because deal announcements typically focus on price and strategic significance. But if the target has factories, employees, customers and an after-sales network in China, those operations are both assets and a basis for Chinese jurisdiction, two sides of the same coin. Moving the business from ABB’s balance sheet to SoftBank’s does not relocate its Chinese entities. They remain where they are, only under a new ultimate owner.

The original report summed this up in a single phrase: the business has “extensive operations in China.” Those few words carried enormous weight in determining whether the deal could proceed.

Security Review and Antitrust Review Are Two Different Gates

When M&A approvals come up, most people first think of antitrust scrutiny. But this deal went through a different gate.

Antitrust reviews focus on competition: whether combining two companies would create enough market power to manipulate prices or squeeze rivals. They rely on defined calculations, allow room for companies to make their case and generally follow public review timelines. Security reviews ask a different question: whether a transaction could affect national security. That standard is far broader than market share, and the process is not public. Outsiders often see only the outcome.

The agency with the long name in ABB’s statement, the Office of the Working Mechanism for the Security Review of Foreign Investment under the National Development and Reform Commission, oversees that second gate. Its job is not to decide whether a deal makes financial sense, but whether it may proceed at all.

Notably, ABB’s statement said only that the deal had been blocked, without specifying the grounds. The NDRC has not responded publicly either. So far, the only fact confirmed by public information is the outcome, not the rationale. That is not unusual in such reviews: withholding the reasons is itself part of the institutional design.

Why 7,000 Employees and BMW’s Production Lines Matter

The $5.4 billion price must be understood in the context of how the industrial robotics business works.

Industrial robots are not off-the-shelf products. The same robotic arm installed on welding lines for different car models will face different cycle-time requirements, workstation constraints, fixtures and machine-vision setups. Selling the equipment is only the beginning. Installation, commissioning, process adaptation and maintenance account for much of the business that follows. Once a customer entrusts a production line to one brand, switching suppliers becomes prohibitively expensive. It means rebuilding the process, not merely replacing a machine.

That is the real significance of the division’s 7,000 employees and its role supplying robotic arms to automakers such as BMW. SoftBank was not buying a product. It was buying a business that has operated on real production lines for years and is embedded in customers’ manufacturing processes. Building something comparable from scratch might not be possible even in a decade.

What Masayoshi Son Wanted

The original report offered one assessment: Son saw the deal as a key part of his push to expand SoftBank’s influence in AI infrastructure.

That point deserves unpacking. SoftBank is an investment group whose strength lies in identifying targets, placing bets and waiting for valuations to rise, not in manufacturing. Robotics is precisely where that investment playbook works least well. The business requires production lines, years of process expertise and teams of engineers who can work on-site. These are slow-moving capabilities that money alone cannot instantly create. Buying a mature business would effectively allow SoftBank to exchange cash for time.

Another part of the backdrop is OpenAI. The original report noted that the acquisition had been widely expected to boost SoftBank’s revenue and profit while it waited for a lift from an eventual OpenAI public listing. In other words, the deal was more than a strategic move on SoftBank’s books. It would also have brought in revenue that could be consolidated into its financial statements. To sell the physical AI story to capital markets, SoftBank needs more than unrealized gains on investments to support it.

That source of revenue has now disappeared, forcing SoftBank to find a different way to tell the story.

After the 7% Drop

SoftBank shares fell about 7% in Tokyo on Thursday, a clear negative response. The market had already factored the deal into SoftBank’s physical AI ambitions and must now remove those expectations.

The SoftBank spokesperson’s decision to decline comment suggests the matter may not yet be over. In cross-border security reviews, companies have previously responded to rejection by refiling, restructuring a transaction or carving out sensitive assets before returning to negotiations. Neither ABB nor SoftBank has said the deal has been terminated.

Only one thing is certain: before this $5.4 billion deal can close, it must first clear China’s review. Public information offers no answer as to why it failed to do so, or whether a different structure could still secure approval.