Is the era of the borderless AI startup over? Yes, but only if you are trying to sell to the wrong side of the fence.
The recent drama surrounding Manus is a textbook example of how geopolitical borders have finally caught up with the latent space. For those not tracking the whispers, Meta tried to secure a $2 billion deal for the AI agent startup. Then Beijing stepped in and told Meta to pack its bags. Now, in a move that feels less like a market transaction and more like a state-mandated hand-off, Tencent is in talks to buy a majority stake in Manus at that same $2 billion valuation.
It is a clean swap. Meta is out, Tencent is in, and the valuation remains frozen in time as if the last six months of AI volatility never happened. According to The Decoder, Tencent is looking to fold Manus’s agent capabilities into its own ecosystem, with a very specific eye on WeChat (which is basically the internet for half the planet).
The irony here is delicious. Meta spent the capital and did the due diligence, only to act as the unpaid scout for Tencent. It is like a sports team scouting a star player for years, negotiating a contract, and then having the league office force them to trade that player to their biggest rival for nothing more than the satisfaction of following the rules.
(I suspect the Meta team is not feeling particularly satisfied).
This isn’t about “synergy” or “strategic alignment.” It is about the hard reality of sovereign AI. If you are building agentic workflows that can actually execute tasks—not just chat, but move pixels and click buttons—you are building a tool for systemic control. Beijing isn’t about to let that infrastructure be owned by a Menlo Park entity, regardless of how many billions are on the table.
The real question is whether Tencent actually wants Manus, or if they are just filling a quota. Tencent has its own internal AI labs and a massive appetite for agentic automation. But integrating a third-party startup into the monolithic architecture of WeChat is a special kind of hell. We have seen this before with the way Tencent absorbs companies; they either strip the talent and kill the product or let it wither in a silo.
Still, the move makes sense if you view AI agents as the new UI. If Tencent can successfully pivot WeChat from a messaging app to an agent-driven operating system, they don’t just own the communication channel—they own the action layer.
But here is the contrarian read: this deal might be a poisoned chalice for Manus. Startups usually thrive on the agility of their founders and the freedom to pivot. Being absorbed by a state-aligned giant like Tencent usually means the “innovation” part ends and the “compliance” part begins. The talent at Manus likely wanted the Meta exit for the prestige and the global reach, not to become a feature in a closed garden.
It’s a political move dressed as a venture deal.
Whether the technical integration actually works is a different story. The friction of merging different agentic frameworks often leads to a bloated, lagging experience that satisfies a slide deck but frustrates a user. If the latency on these agents is high, the whole “agentic OS” dream dies in the first five seconds of a user’s request.
Tencent will have a Manus-powered agent live in a beta version of WeChat by Q4.
If they pull it off, they’ve successfully weaponized a regulatory blockade to leapfrog Meta’s agent strategy in the East. If they don’t, they’ve just paid $2 billion for a very expensive set of lessons on why you shouldn’t buy a company that was already rejected by the government.
