In April 2026, the world witnessed something that would have seemed impossible just a few years ago: a government ordering two companies to undo a deal that was already done.

‍Meta had acquired Manus AI, a Singapore-registered AI startup, for up to $3 billion in December 2025. The product was live. Integration had begun. The press releases had gone out. And then Beijing said: unwind it.

China's National Development and Reform Commission (NDRC) didn't merely block a pending merger. It reached across borders, into a completed transaction involving a Singapore-based entity and an American tech giant, and ordered them to reverse course. It also barred Manus's co-founders from leaving China. This was not a routine antitrust review. This was a declaration: that geography on paper does not equal sovereignty in practice.

For founders building companies at the frontier of technology — especially artificial intelligence — the Manus story is not a cautionary tale from someone else's industry. It is the clearest possible signal that geopolitics has become a foundational business risk, not a footnote in the legal section of your pitch deck.

Beijing's Long History of Keeping Its Founders Close

To understand Manus, you first need to understand the environment that produced it. China's relationship with its tech founders has never been purely economic. It has always been political. The crackdown that began in late 2020 wiped more than a combined $1 trillion from the valuations of China's biggest technology companies — and it was triggered not by market failure, but by a speech.

In October 2020, Alibaba founder Jack Ma publicly criticized China's financial regulators. Within days, the government blocked the $37 billion IPO of Ant Group, Alibaba's fintech arm — at the time, what would have been the largest public offering in history. Ma effectively vanished from public life for months. The message to every founder in China was immediate and unambiguous: the state is the senior partner, always.

Didi, China's dominant ride-hailing company, found this out the hard way when it proceeded with its New York IPO in June 2021 against Beijing's wishes. Chinese regulators launched a cybersecurity probe within days of the listing, eventually forcing Didi to delist from the NYSE and retreat to Hong Kong. The company's valuation fell from $70 billion at IPO to a fraction of that figure.

These actions followed a consistent logic: Chinese companies, regardless of where they list or where they register their holding companies, remain within the orbit of the Chinese state. The data they hold, the talent they employ, the research they generate — all of it is subject to Beijing's claim. The crackdown was, as analysts noted at the time, less about consumer protection and competition policy than about one word: control.

Platform companies were forced to issue "golden shares" to state-owned enterprises — minority equity stakes that carried disproportionate governance rights, effectively giving the state veto power over strategic corporate decisions. The era of the freewheeling Chinese tech entrepreneur had ended.

The Rise of Manus: China's Second DeepSeek Moment

Against this backdrop, Manus AI arrived in March 2025 like a thunderclap. Developed by Butterfly Effect, a startup founded by Xiao Hong — a 2015 graduate of Huazhong University of Science and Technology — and co-founder Ji Yichao, Manus launched in invitation-only beta on March 6, 2025.

The launch demo video, which showed the agent autonomously completing tasks including resume screening and real estate analysis, drew more than one million views within twenty hours. Demand for invitation codes became so intense that codes were reportedly resold on Chinese platforms for the equivalent of $7,000 to $13,800 each.

Tech observers quickly drew parallels with DeepSeek, the AI model that had rattled Silicon Valley just months earlier. "This AI agent called Manus is going crazy viral in China right now," wrote Rowan Cheung, founder of The Rundown AI newsletter. "It's like Deep Research + Operator + Claude Computer combined, and it's REALLY good." Some dubbed it "China's second DeepSeek moment", and Chinese state media declared Manus "the next DeepSeek" shortly after its launch.

What made Manus distinctive was its architecture: unlike DeepSeek's single large language model, Manus used multiple AI models — including Anthropic's Claude 3.5 Sonnet and fine-tuned versions of Alibaba's Qwen — coordinating as independent agents to execute complex, multi-step tasks autonomously. It claimed state-of-the-art results on the GAIA benchmark for real-world task completion.

The company had earlier declined acquisition offers. In 2024, ByteDance approached Butterfly Effect with a $30 million offer, which Xiao Hong turned down. By April 2025, after a Series B round led by Benchmark, the company was valued at approximately $500 million — a figure that would multiply several times over before the year was out.

The Geography Play: Singapore and the 'China Shedding' Playbook

Manus had global ambitions from the start. As co-founder Xiao Hong noted bluntly in a podcast: "You can price in USD, and with the exchange rate that's a sevenfold multiplier. Even if we're only operating at 10% power because of cultural differences overseas, we'll still make more than in China."

The company had also declined local government investment offers from several Chinese municipalities, citing concerns that domestic government ties could create scrutiny in Western markets and complicate its global business. The founders understood the calculus: Chinese government money came with strings attached, and Western investors — especially American VCs — were becoming increasingly cautious about portfolio exposure to Beijing.

In mid-2025, Butterfly Effect relocated its headquarters from Beijing and Wuhan to Singapore. Three co-founders relocated. Around 40 of roughly 120 China-based technical staff moved with them. The rest were laid off. The company closed its Chinese-language social media accounts, blocked access from mainland China, shelved a planned Alibaba partnership, and began operating exclusively under a Singapore-registered entity.

This was a playbook that had become familiar in Chinese tech circles — what some were calling 'China shedding': the strategic geographic restructuring of a company to distance it from its Chinese origins, reduce regulatory exposure, and make it more palatable to Western investors and acquirers. The logic was sound on paper. Singapore had a stable regulatory environment, close proximity to Asian capital markets, and a credible posture of independence from Beijing's regulatory orbit.

But the strategy carried an implicit assumption: that a change of address would constitute a change of jurisdiction. Meta announced its acquisition in December 2025, with plans to integrate Manus's agentic AI technology into Meta AI to compete with Google and OpenAI. The deal was reportedly valued at between $2 and $3 billion. Manus's website was updated to read: "Manus is now part of Meta."

Beijing's Counter-Move: The Crackdown That Changed Everything

China's response was swift and unprecedented. In January 2026, China's Ministry of Commerce announced a formal investigation into whether the acquisition complied with China's export control laws, technology transfer regulations, and foreign investment rules. The probe elevated quickly: according to reports, the decision was ultimately referred to China's National Security Commission — the Communist Party body chaired by Xi Jinping himself.

On April 28, 2026, the NDRC issued its ruling: the acquisition must be unwound. The statement was brief but emphatic — the decision had been made "in accordance with laws and regulations," and all parties were instructed to withdraw the acquisition transaction. The co-founders of Manus, Xiao Hong and Ji Yichao, were barred from leaving China — a standard Chinese tool in investigations, but one that underlined how personal the state's interest in this transaction had become.

Beijing's state-run Global Times was direct about the reasoning: "Manus's early R&D was conducted in China and its core data originated there. The key issue is not where the company is registered or where its team is currently based. It lies in the extent of its technological, talent and data links with China, and whether the transaction could harm China's industrial security."

Meta stated the transaction "complied fully with applicable law." The White House backed Meta. But Meta — which derives roughly 11% of its revenue from China-based advertisers — ultimately indicated it would comply. The deal that was already done was, for now, undone.

The ruling was, as analysts noted, the first time China had deployed its foreign investment security review mechanism — introduced in late 2020 but rarely used — to block a post-completion transaction. The precedent was stark: there is no safe harbor in a completed deal. If Beijing considers your technology strategically important, the transaction can be reversed at any point.

What This Means for Founders: Navigating Geography as Strategy

"Clearly after Manusgate, founders will know that if you start in China, you stay in China," said Duncan Clark, chairman of consultancy BDA China and an early advisor to Alibaba. The statement captures the new reality bluntly.

For founders building in or near China — or indeed anywhere geopolitical fault lines run deep — the Manus saga has rewritten the risk calculus in several important ways.

First: incorporation jurisdiction is not the same as regulatory jurisdiction. Manus was legally a Singapore company. Its founders had physically relocated. It had laid off its China-based staff, closed its Chinese social media accounts, and scrubbed its mainland presence. None of that mattered. What mattered was where the technology was researched, where the data originated, and where the talent had been trained. Governments are increasingly applying extraterritorial logic to technology assets.

Second: the geographic arbitrage play has limits. 'China shedding' — the relocation of Chinese-origin companies to Singapore, Dubai, or other neutral hubs — was a rational response to a genuinely difficult regulatory environment. But as the Manus case demonstrates, it works only if the underlying technology and talent have genuinely moved, not just the registered address. Founders considering this approach need legal and geopolitical counsel, not just incorporation advice.

Third: both Washington and Beijing are tightening the screws simultaneously. Bloomberg reported that Beijing is considering rules requiring Chinese AI companies to obtain approval before accepting U.S. investment. Meanwhile, the U.S. Treasury's Outbound Investment Security Program already requires notification of investments in AI companies in "countries of concern." Founders and their investors are now caught in a regulatory crossfire from both sides.

Fourth: timing a deal around geopolitics is no longer sufficient. The Manus acquisition was announced weeks before a planned Trump-Xi summit — a window that might once have offered diplomatic cover. Instead, the deal became leverage in a broader negotiation, a bargaining chip in a trade and technology standoff that neither side was willing to concede on.

The practical upshot for founders is this: geography is no longer a logistical decision. It is a strategic one, with permanent consequences for your exit options, investor pool, acquirer universe, and regulatory exposure. The question of where you incorporate, where your servers sit, where your engineers are based, and where your data flows is now as consequential as your product roadmap.

The Broader Signal

The Manus story is not ultimately about one company or one deal. It is about the end of a comfortable fiction: that the global technology economy was a geography-free zone where capital, code, and talent could move freely across borders, answerable to market logic alone. As one analysis put it, "simply shifting corporate registration offshore does not place a company beyond China's extraterritorial control and regulatory reach if its technology, founders, or research ecosystem remain tied to the mainland."

For founders operating in the AI era, this is the landscape: two superpowers competing fiercely for technological advantage, each deploying regulatory, investment, and export control tools as instruments of national strategy. The idea that a startup can remain apolitical — that a great product and smart investors are sufficient insulation — is an increasingly expensive illusion.

The founders who will navigate this era successfully are those who treat geopolitical risk as a design constraint from day one: choosing their jurisdictions intentionally, structuring their cap tables with regulatory exposure in mind, and building contingency plans for the possibility that their most important deal might be blocked — or reversed — by a government that sees their technology not as a product, but as a strategic asset.

Manus built something remarkable. The question the next generation of founders must answer is not just what they build — but where, and for whom. In the age of AI statecraft, the map is the message.

‍

Read - This Shop Is Run Completely by an AI Agent: Let's Walk You Through What the Future of Shopping Looks Like