In a move that reads like a techno-thriller screenplay, China's National Development and Reform Commission (NDRC) has reportedly imposed an exit ban on Manus AI co-founders Xiao Hong and Ji Yichao, effectively grounding them while regulators scrutinize Meta's proposed $2 billion acquisition of the AI startup.

The decision, first reported by Chinese state media outlet Caixin on March 25, marks one of the most dramatic escalations in the ongoing geopolitical tug-of-war over artificial intelligence technology. While Manus AI is technically incorporated in Singapore, its core research team and foundational technology were developed in Shenzhen — a detail that Beijing is now leveraging to assert jurisdiction over the deal.

What Happened

Meta announced its intention to acquire Manus AI in early March 2026, valuing the company at approximately $2 billion. Manus had gained international attention for its autonomous AI agent platform, which demonstrated remarkable capabilities in executing complex multi-step tasks with minimal human intervention.

The acquisition seemed like a natural fit for Meta's aggressive AI strategy — the company has already committed over $115 billion in AI infrastructure spending for 2026 alone. But Beijing had other plans. Within days of the deal's announcement, the NDRC initiated a formal review under China's outbound technology transfer regulations, arguing that Manus AI's core intellectual property constitutes "strategic technology" subject to export controls.

The exit ban on Xiao and Ji is a particularly aggressive measure. Sources familiar with the matter say the founders were notified on March 22 that they would not be permitted to leave mainland China until the review is complete — a process that could take anywhere from three to twelve months.

Why This Matters

This isn't just about one deal. The Manus AI situation has become a litmus test for how China plans to handle AI technology transfers in an era of escalating US-China tech competition. Several key dynamics are at play:

The Singapore Shell Problem: Manus AI's Singapore incorporation was widely seen as a strategy to distance itself from Chinese regulatory oversight. Beijing's willingness to look through the corporate structure and assert control based on where the technology was actually developed sends a clear message to other Chinese AI startups contemplating similar arrangements.

The Talent Trap: By restricting the founders' movement, China is effectively using human capital as leverage. This tactic has been employed before in financial fraud cases, but applying it to a technology transfer review is a significant escalation that will make every AI researcher in China think twice about international deals.

Meta's AI Ambitions at Risk: For Meta, this comes at an awkward time. The company just laid off 700 employees while simultaneously unveiling a stock compensation program that could pay top executives up to $921 million each. Acquiring Manus was supposed to bolster Meta's autonomous agent capabilities — a key strategic priority as the company races against Google and OpenAI in the agentic AI space.

The Bigger Picture

The US and China are locked in what many analysts are calling a "cold war for AI supremacy." Washington has imposed increasingly strict semiconductor export controls, while Beijing has responded with its own restrictions on critical mineral exports and now, apparently, human talent.

Manus AI sits right at the fault line. Its agent technology represents exactly the kind of capability that both superpowers want to control — autonomous systems that can plan, reason, and execute complex tasks across digital environments. The fact that this technology was developed in China but packaged in a Singapore entity creates a jurisdictional gray zone that both sides are now rushing to claim.

My Opinion

Look, I'll be real with you — this story is wild, and it perfectly encapsulates everything that's broken about the current state of AI geopolitics.

On one hand, you have China essentially holding two entrepreneurs hostage in their own country because they built something valuable. That's terrifying, and it should be a wake-up call for every founder operating in this space. Your exit strategy isn't just a slide in your pitch deck anymore — it might literally determine whether you can leave the country.

On the other hand, let's not pretend Meta is the innocent party here. This is a company that just fired 700 people while preparing to hand their executives nearly a billion dollars each in stock options. They're trying to buy their way to AI dominance because they can't build it fast enough in-house, and they're shocked — shocked — that geopolitics got in the way.

The real losers here? The engineers and researchers caught in the middle. People who just wanted to build cool AI agents and are now pawns in a chess game between Mark Zuckerberg and the Chinese Communist Party. Welcome to 2026, where your GitHub commits might determine your immigration status.

My prediction: this deal will eventually collapse. China won't approve it, Meta will move on to the next acquisition target, and Manus AI will be left in limbo — too controversial for Western buyers, too restricted by Beijing to operate freely. It's a lose-lose-lose situation, and the only people who win are the lawyers and regulators who get to bill by the hour while two brilliant founders sit in Shenzhen, watching their life's work become a diplomatic bargaining chip.

If you're a Chinese AI founder reading this: get a really good lawyer. And maybe a second passport.