Inside the Meta Manus deal and why it mattered
Meta Manus deal discussions began in December, when the social media giant agreed to buy Manus, a “general-purpose” AI agent developed by Butterfly Effect. Manus, founded in China but now based in Singapore, builds autonomous agents that can execute multi-step tasks such as sifting and summarising résumés or creating a stock-analysis website. Meta framed the acquisition as a way to “bring a leading agent to billions of people and unlock opportunities for businesses” across its platforms, and analysts suggested the transaction could be worth more than USD 2 billion (approx. RM9.2 billion). For Meta, Manus promised a shortcut to richer AI agent capabilities that could be embedded into Facebook, Instagram and messaging products. For the industry, it was a rare example of a major US platform buying an AI startup with deep Chinese roots, testing how far cross border AI investment would be tolerated amid intensifying tech rivalry.
How China blocked the acquisition and what regulators said
The deal ran into a wall when China’s National Development and Reform Commission (NDRC), acting through its Office of the Working Mechanism for Security Review of Foreign Investment, prohibited the foreign acquisition of Manus and ordered all parties to withdraw from the transaction. The NDRC statement, issued after an earlier decision to review the deal, said the move was taken in accordance with laws and regulations but did not name Meta directly or spell out detailed reasons. A separate account linked the intervention to national security regulations for foreign investment that have been in force since 2021 and emphasised concerns over technology and data security ties to China, rather than Manus’ Singapore registration. Chinese authorities also reportedly restricted Manus co-founders Xiao Hong and Ji Yichao from leaving the country during the review, underscoring how seriously Beijing now treats outbound transfers of advanced AI capabilities and talent.
From one blocked deal to a broader AI security red line
China blocks AI acquisition headlines might be easy to view as a single corporate setback, but the Meta Manus deal has quickly become a symbol of a deeper policy shift. Regulators have signalled that incorporation abroad is not enough to sever what they see as sensitive ties in technology, talent and data. Manus was recognised as a leading AI innovator built with Chinese engineers and infrastructure, even though it did not create its own foundation models, and that lineage appears to have triggered security concerns. Beijing has made clear it does not want domestic AI know-how flowing unchecked to major US platforms, placing AI startup regulation squarely in the realm of national security. For founders, the episode highlights that abrupt separation from Chinese operations after taking foreign capital may no longer be viable, especially when strategic technologies and cross-border AI investment are involved.
Rising risk for global AI investors and changing deal playbooks
The decision to unwind Meta’s move has rattled investors exposed to global AI M&A involving firms with Chinese roots. Legal experts now warn that simply relocating a startup’s headquarters or registering IP offshore will not shield it if authorities view its technology as strategically important. Investors in Chinese-founded businesses may need to insist on more substantial operational separation: localising intellectual property, relocating research and development, and ring‑fencing China-facing data flows before any sale or merger. For big tech buyers, strategic responses could include favouring minority stakes, joint ventures with on‑shore R&D centres, or licensing arrangements that avoid outright control of sensitive assets. These structures may help navigate tightening review regimes while still enabling collaboration. At a minimum, the Meta Manus deal saga elevates political and regulatory risk to the same level as valuation and technology fit in cross border AI investment decisions.
AI competition, industrial policy and the DeepSeek backdrop
The blocked Meta Manus deal lands amid intensifying competition in AI, where industrial policy and market dynamics are increasingly intertwined. Chinese model developer DeepSeek has repeatedly disrupted expectations by delivering strong models using fewer computing resources than leading US systems, challenging assumptions about the impact of export controls. Its latest DeepSeek-V4 and DeepSeek-V4 Pro releases, while drawing a more muted market reaction, sit in a fiercely competitive field of open‑weight models. To win share, DeepSeek has offered steep discounts on access to DeepSeek-V4-Pro, a move that threatens to restart a price war and lowers barriers for developers and startups. At the same time, Beijing’s clampdown on outbound AI acquisitions shows it is intent on keeping such capabilities within its ecosystem. Together, aggressive domestic competition and tight outbound controls illustrate how states are using both market incentives and national security tools to shape the trajectory of global AI M&A.
