Meta announced its acquisition of Manus in December, and on April 27 China’s National Development and Reform Commission ordered both parties to “withdraw the acquisition transaction” (CNBC, Bloomberg).
The interesting question is not whether Beijing can issue the order. It is what “withdraw” means after a transaction has already moved into operational integration.
What the consensus says
The consensus view is straightforward: Meta tried to secure Chinese-origin AI capabilities via a Singapore structure, Beijing applied look-through jurisdiction, and Beijing blocked the deal. That framing fits the legal architecture established by China’s foreign investment security review regime, jointly issued by NDRC and MOFCOM and effective since 2021 (NDRC/MOFCOM Measures PDF).
What this framing gets right is mechanism. What it may miss is timing.
What the timing shows
By the time of the April 27 order, reporting indicated employees had already moved into Meta’s Singapore operations and parts of the integration had already begun (CNN). If accurate, Beijing issued an unwind directive after substantial operational migration had already occurred.
The order also landed shortly before the May 14-15 Trump-Xi summit in Beijing, a meeting publicly described as postponed from an earlier slot amid Iran conflict management and Hormuz disruption (Reuters, BBC).
There is a doctrine-versus-practice tension. If Beijing’s main objective were blanket prevention of U.S. access to Chinese-origin model capability, enforcement would likely extend beyond M&A into broad limits on open model distribution. Yet major Chinese model families, including DeepSeek R1 and Alibaba’s Qwen line, have been released as open artifacts that can be downloaded and run globally (DeepSeek GitHub, Alibaba Qwen3 release). That does not invalidate the Manus intervention, but it complicates a pure technology-containment reading.
The working hypothesis: The NDRC order functions less as a practical unwind mechanism and more as a durable bargaining instrument: an unresolved compliance liability that can be tightened, relaxed, or resolved in a broader negotiation.
Why the unwind may be structurally hard
Unwinding a cross-border deal after integration has begun is not the same as blocking it pre-close. It typically requires synchronized reversals across capital flows, personnel, systems, and governance structures. That process is legally and operationally harder than issuing the initial order. The practical result is a long-lived compliance overhang that can be calibrated over time.
The energy context
The Strait of Hormuz remains one of the world’s most important energy chokepoints, with roughly one-fifth of global petroleum liquids consumption and about one-fifth of global LNG trade moving through it (EIA, EIA LNG).
China is central to that flow: the IEA notes that China and India together receive 44% of crude exports that transit Hormuz (IEA). In that setting, summit timing and energy bargaining context matter for interpreting policy moves around unrelated but politically useful files.
Against that backdrop, the Manus order is a relatively low-cost lever if its primary function is to manufacture compliance uncertainty rather than recover assets.
What would change my mind
-
No movement after the summit window. If the Manus compliance posture remains unchanged 30 days after the May 14-15 summit and no adjacent concession appears, this looks more like durable doctrine than traded leverage.
-
A genuine unwind actually occurs. If capital, personnel, and IP separation happen in a way that satisfies the order’s practical requirements, the “impossible-by-design” framing is wrong.
-
Beijing formalizes broad guidance beyond Manus. Explicit doctrine extending this enforcement posture across similarly structured Chinese-origin AI transactions would shift the base case toward precedent-building over summit bargaining.
Related: Investor Companion: The Deal That Can’t Be Undone — market implications and monitored signals for this same thesis.
If you found this useful, the best thing you can do is forward it to one person who would push back on it. I’d rather be wrong in public than right in private.