Key Takeaways

  • Manus’s early investors are preparing a $2 billion buyback from Meta after Chinese regulators pushed to unwind the original deal
  • The reversal highlights intensifying government scrutiny of cross-border AI assets and ownership
  • Manus’s rapid revenue growth has increased investor urgency to regain control despite ongoing regulatory uncertainty

The unusual situation unfolding around Manus has become one of the more revealing episodes in how governments are starting to intervene in the ownership of frontier AI companies. Original backers of the Singapore-based startup, including HSG, ZhenFund, and Tencent, are now moving to repurchase the company from Meta at the same $2 billion valuation Meta paid in late 2025. The plan follows direct pressure from Chinese authorities, which ordered the unwinding of Meta’s acquisition due to concerns tied to foreign investment and technology transfer. Beijing’s review shows how national-security considerations can override commercial M&A logic, particularly in sensitive AI categories.

This is not a distressed asset sale or a valuation reset. The investors are reportedly willing to buy Manus back at full price. The Information reported that Manus’s annualized revenue run rate is now between $400 million and $500 million, up from about $100 million earlier in its growth cycle. Such a rapid revenue increase in a short window helps explain why early shareholders are motivated to regain control.

According to global funding figures cited in the Stanford HAI AI Index 2024, overall AI investment reached $325.9 billion in 2023, and generative AI alone attracted $25.2 billion. That scale of capital flow heightens demand for cross-border AI assets like Manus and draws regulators deeper into deal oversight, especially as models approach more autonomous capabilities. The policy environment in the United States similarly reflects this trend, where the Executive Order on AI and follow-on actions have encouraged additional focus on model risk, export controls, and safety assessments. NIST placed particular emphasis on these areas through the release of the AI RMF 1.0, which enterprises increasingly reference as a baseline.

Pressure from Beijing began to build soon after Meta announced its acquisition. China cited national-security concerns and limits on foreign investment in strategic AI technologies. According to CNBC, regulators instructed Meta to unwind the deal, setting in motion a process that is now culminating in the buyback discussions reported on June 18, 2026. Meta has already started decoupling the two companies, including a halt to data-sharing arrangements. Some reports suggest Manus could later be reorganized as a China-based joint venture, eventually opening the door to a Hong Kong listing.

Manus competes directly in a rapidly evolving market against developers like OpenAI, Anthropic, and Mistral. These firms are racing to build agentic AI systems that do more than generate text. They are training models capable of orchestrating tasks, navigating workflows, and in some cases interacting with software autonomously. Cross-border ownership of companies developing such systems often draws regulatory attention as governments monitor data sovereignty, military applications, and control over advanced model weights.

Meta continues to spend heavily on infrastructure and model development while navigating international regulatory challenges. Some analysts, such as those at Reuters, noted that Meta may eventually decide it is better to exit the Manus relationship cleanly rather than continue negotiating with regulators.

A joint venture inside China could set the foundation for local scaling, but executing that plan requires approvals and cooperation across multiple agencies. Operating under its original investors, including Tencent, ZhenFund, and HSG, may help Manus maintain product velocity by leveraging their experience navigating China’s regulatory environment. It might also help the company align with the country’s AI development priorities, which increasingly revolve around industrial automation and secure compute.

As enterprises evaluate AI suppliers, they are paying closer attention to model lineage and ownership. Several advisory firms, including Deloitte, have noted that supply-chain clarity heavily influences AI adoption decisions. If Manus transitions back to its previous investor group and reshapes its corporate structure around a China-based joint venture, enterprise buyers outside China will need to reassess data handling, export controls, and long-term support expectations. Those questions are becoming common as buyers map vendor footprints against the NIST AI RMF 1.0 and ISO/IEC 42001 guidelines.

The timing of the revenue surge means Manus has gained leverage in deal negotiations. Growing from roughly $100 million to as much as $500 million in annualized run rate in a short period strengthens investor confidence and suggests the company has found a durable market for its agentic systems. How that momentum holds up during a forced ownership transition remains an open question, but the growth trajectory provides a financial cushion.

Manus illustrates the new realities shaping cross-border AI investment. Regulatory intervention can quickly reshape deal economics, forcing investors to plan for revisions even after agreements close. Governments are increasingly intervening when stakes touch on national security or strategic technology leadership, shifting global AI competition from pure commercial rivalry to intense geopolitical scrutiny.