Hello everyone, I’m Uncle Guo.
The more I look at the Manus situation, the more I feel it was perfectly positioned to serve as the “chicken killed to scare the monkeys.”
That sounds a bit harsh, but that’s how the business world often works. When your turn comes, it’s not necessarily because you’re the worst, or because you’ve committed an unforgivable sin. It’s because you were just big enough, loud enough, and flashy enough—but not so big that you were “too big to fail.” You were standing in a spot that made you perfectly visible, and as a result, everyone can learn a lesson from your misfortune.
Manus fits this description perfectly.
A $2 billion valuation, a buyer like Meta (one of the “Big Seven”), a Singaporean corporate entity, a Chinese founding team, AI Agents, global tech expansion, and a capital exit. Put these keywords together, and you don’t even need a PowerPoint deck; the story itself is a financing pitch that is almost too beautiful to be true.
That is exactly where the problem lies.
It looked so much like a success story that someone had to step in and provide a reminder: this path isn’t as easy as it looks.
Don’t Write This Off as Just Another Failed AI Expansion
When people talk about Manus, they naturally drift toward two extremes. One is nationalistic sentiment, claiming that Chinese regulators finally blocked Meta from acquiring Chinese AI assets. The other is product criticism, arguing that Manus wasn’t that impressive to begin with, the “Agent” narrative was overblown, and a fall was inevitable.
Both angles are valid, but both are superficial.
What makes Manus truly interesting is how it tangled together several threads that are usually viewed separately: capital exits, technology origins, corporate structure, talent mobility, data permissions, and the valuation stories that AI application companies love to tell. It acts like an X-ray, exposing things in the domestic AI venture capital industry that aren’t usually discussed in public.
Over the past two years, many AI entrepreneurs have kept a “secret ledger.” Raising RMB funds domestically isn’t easy, and getting listed on the A-share main board is far from guaranteed. For AI application companies, spinning a beautiful capital story domestically is a high-stakes challenge. The Hong Kong stock market is more realistic, but it demands revenue, profit, growth, compliance, and market sentiment—it won’t automatically grant you a higher valuation just because you have “AI” in your name.
So, many have settled on a different path: set up the structure abroad, target the global market, raise USD funding, and if you can eventually sell to a major US tech giant—ideally someone like Meta, Google, or Microsoft—then you’ve “won the game.”
Manus almost turned this script into a blueprint.
That is exactly why it was so dangerous.
The Manus Script Was Too Perfect—Almost Blindingly So
If you break down the Manus trajectory, you’ll find it checked every box that the AI startup scene has been craving over the past year.
First, the viral explosion. A “General Agent” concept paired with flashy demos, benchmarks, invite-only access, and social media buzz—the hype was instant. Then came the funding; when a name like Benchmark appears, people automatically assume the company has been “vetted by Silicon Valley.” Then came the revenue narrative: ARR, global users, paid growth, productivity tools. These terms are familiar to investors; they don’t necessarily prove a company is healthy, but they are easy to sell.
Finally, the Meta acquisition.
This is the most thrilling part of the script. If you are a domestic AI entrepreneur, it’s hard not to be tempted. A Chinese team builds the product, a Singaporean entity handles globalization, USD funds flow in, a US giant acquires it, and the team and investors exit. The entire story—from the startup narrative to the capital narrative to the globalization narrative—ran as smoothly as if it were on a VIP track.
But that smoothness is, in itself, the problem. It was too smooth, too complete, and too much like a success template that others could copy.
So, it had to be made less smooth.
It Was Big Enough, but Not “Too Big to Fail”
Why Manus, specifically?
A very practical reason is that it was big enough. A $2 billion deal is no longer small potatoes for an AI startup. The buyer was Meta, not some obscure fund. Once this case was brought to light, the message was guaranteed to spread.
To “kill the chicken to scare the monkeys,” the chicken can’t be too small. If you shut down a small team with a few million dollars in funding, people will just think it’s an isolated incident; the story won’t travel far, and the deterrent effect will be weak. Manus was different. It had viral value, capital symbolism, and industry labels, and it allowed many people to project themselves into the situation.
Yet, on the other hand, it wasn’t so big that it was untouchable.
It wasn’t NVIDIA, OpenAI, Meta, ByteDance, Apple, or Google—companies that control the traffic gateways and infrastructure. At its core, it was still an application-layer Agent company. Today, browser agents, local PC agents, coding agents, and enterprise workflow agents are popping up everywhere. More realistically, the most comfortable form of an agent will likely be closer to your computer, system, account environment, and real-world workflow, rather than waiting for you to click on a web page.
A general-purpose agent living on the web certainly has value, but not so much that it’s untouchable. This position is awkward: it’s loud enough to be a public example, but not “hard” enough that destroying it would break the industry’s legs.
In the business world, this is the most dangerous position to be in.
The Real Taboo: Redefining “Going Global”
The most troublesome part of the Manus case isn’t necessarily that they did something heinous. At least based on public information, we can’t draw that conclusion.
The real trouble is that it was too easy to slap a label on them that would be impossible to peel off.
When an entrepreneur looks at Manus, they see globalization. A Singaporean entity, USD funding, overseas markets, acquisition by a US giant—how beautiful. But look at it from another perspective, and the picture changes: the talent has left, the technology has left, the assets have left, and the capital gains are leaving too.
You call it “AI going global,” but others might wonder: is this industrial expansion, or is it the migration of technology and capital?
That’s a harsh sentiment, but it’s the shadow that hangs over the Manus case. If a company just sells products overseas and brings back USD, everyone is happy—it’s good for foreign exchange, growth, and branding. But if a company starts with an offshore structure, raises money offshore, keeps its entity offshore, and finally exits to an offshore buyer—especially a US tech giant—the flavor changes.
You explain that it’s globalization; others interpret it as capital flight. You explain it as a commercial exit; others interpret it as a blueprint for others to follow. You explain it as technology export; others see only four words: “taking the whole pot.”
In today’s environment, once some labels are attached, the cost of explanation is absurdly high. You can write legal opinions, provide compliance statements, and argue that the corporate structure and transaction process are legal, but sometimes the problem isn’t how you explain it.
The problem is that others no longer want this story to remain a template.
The Real Victim: The Exit Fantasy of Domestic AI VC
The Manus case might not be the biggest blow to Manus itself.
It hurts the already narrow exit path for the domestic AI venture capital industry.
Domestic AI startups are in an awkward spot. Are there no opportunities? Of course not. China has the scenarios, the engineers, the supply chains, the customers, and a bunch of people willing to grind. But the capital path isn’t that pretty. RMB funds have been cautious in recent years, and for many AI application companies, the A-share market is like climbing a mountain with no stairs. The Hong Kong market is more realistic, but not as easy as imagined. As for being acquired by domestic giants, the price, strategy, synergy, and regulatory hurdles are all heavy burdens.
So, USD funding and overseas M&A once acted like a tempting escape hatch.
After the Manus incident, this channel will be repriced. Foreign capital will be more cautious about investing in Chinese-background AI companies; domestic teams will be more sensitive about offshore structures; acquisitions of Chinese teams by major tech firms will carry an extra layer of uncertainty; and investors looking at exit paths will ask one more question: “Could this become the next Manus?”
Once that question is asked, valuations will drop.
Risk is a funny thing. It doesn’t necessarily have to happen; as soon as it’s written into an investor’s mind, the price has already changed. Therefore, the damage Manus has done to the domestic AI VC industry isn’t just a failed deal; it might make many already fragile stories much harder to sell.
In the past, you could say, “We target the global market, raise USD, and have a chance to be acquired by an overseas giant.” In the future, investors might ask: “And then what? What happens if you get blocked at the final hurdle?”
That’s a buzzkill of a question.
But investing has always been a buzzkill of a business.
Who Wins, Who Loses?
The most interesting part of this is that the winners and losers aren’t entirely on the surface.
First, the biggest winner. Some people win very quietly. A single Manus case might have achieved more than ten official documents ever could. In the future, it will be much harder for domestic AI teams to replicate this route. Regulators will watch, overseas investors will watch, and founders themselves will weigh the risks. The best part is that this doesn’t even need to be discussed daily; the case is there, the amount is there, the buyer is there, and the outcome is there—everyone will learn on their own.
That is the value of a “sample.”
Then there’s Meta. On the surface, Meta is obviously a loser—the deal was blocked, and they certainly had to pay the costs for process, legal, and PR. But did Meta really gain nothing? That’s an interesting question.
When an M&A deal reaches this stage, it’s hard for outsiders to know how many rounds the teams talked, how much technical material was reviewed, how many roadmaps were aligned, and how many core personnel were contacted. The trouble with knowledge is that while money can be returned, equity can be withdrawn, and contracts can be rewritten, how do you “return” the judgments already formed in people’s minds?
I’m not saying Meta necessarily took something they shouldn’t have. Without evidence, I can’t write that. I’m just saying that, based on business common sense, a deal that reaches deep water is unlikely to leave nothing behind.
And honestly, I never thought Meta buying Manus was a stroke of genius. What Meta AI really lacks is probably not a Web Agent team. It lacks something to pull itself out of its “second-tier” narrative—like a model capability that can go head-to-head with Grok, Claude, or Gemini, or a hardware lever that can shift the landscape in compute. Manus is not an answer at that level; this deal looked more like “anxiety-driven procurement.”
A giant realizes its voice is getting quieter at the AI table, sees a hot, well-spoken, globalized Agent company, and wants to buy it just to be safe. Such moves aren’t necessarily stupid, but they aren’t necessarily smart either.
Ordinary employees might not be so miserable, especially those without deep equity ties. Having Manus, a globalized Agent project, and a Meta acquisition attempt on your resume is a layer of gold plating. A company’s fate and an individual’s market value are sometimes not on the same line. That’s the cruelty of startups: the company might be used as a cautionary tale, investors might be uncomfortable, and the founding team might be re-examined, but when ordinary employees change jobs, HR might not even blink at that experience.
The ones who are truly uncomfortable are Manus and its stakeholders. Domestically, the label is heavy; overseas, the risk is too high. The brand value has been punctured, the company value needs to be reassessed, and the investors’ exit path has been repriced.
More troublesome is that Manus isn’t a “normal” failure. A normal failure is about bad products, poor growth, or failed commercialization—people sigh and move on to the next game. The Manus case will become a risk textbook. From now on, whenever an investor looks at a similar company, a thought will flash in their mind: “Could this be the next Manus?”
With that flash, the company’s valuation might drop by a chunk.
Agents Weren’t That Magical Anyway
As an aside, I’ll add this:
What was most attractive about Manus at the time was the “General Agent” narrative. It wasn’t just chatting; it could help you execute tasks, open web pages, organize information, create spreadsheets, run workflows, and even connect to accounts and tools.
This is cool, of course, but as Agents evolve, they look less like pure applications. They start touching permissions, data, accounts, enterprise workflows, customer assets, and, in a sense, accountability for actions. If a chatbot talks nonsense, the worst that happens is a wrong answer; if an agent acts recklessly, accounts could be compromised, data could be stolen, workflows could be disrupted, and customers could be contacted.
Therefore, the valuation logic for application-layer Agents cannot just look at how flashy the demo is. You have to look at where it lives, what permissions it has, what data it processes, who it acts for, and who is responsible when things go wrong.
Manus’s awkwardness lies here, too. On one side, it’s a very sexy product narrative; on the other, it’s a very sensitive boundary of capability. The more capable it is, the more visible the risks; the more globalized it is, the more it will be questioned about its allegiance.
This isn’t just a Manus problem. It’s a problem the entire Agent industry will face sooner or later. It’s just that Manus moved too fast and the spotlight was too bright, so it was the first to be seen.
Finally
At the end of the day, the truly ugly part of the Manus case isn’t that an AI company failed to sell itself; it’s that it shone too bright a light on things in the domestic startup environment that people usually prefer not to talk about.
Why do so many people want to go global? Is it really because they had a vision for the global market from day one? Sure, some do, but not all.
Some realize that after the domestic market is ground down to the bone—where customers demand results, channels demand rebates, platforms demand revenue shares, competitors demand price wars, investors demand growth curves, regulators demand compliance, employees demand stability, and founders have to talk about “long-termism” every day—they feel like they are serving eight tables at once, and they can’t afford to neglect any of them.
So, “going global” becomes a very decent-sounding excuse.
You say “globalization,” and it sounds sophisticated. You say “expanding overseas markets,” and it sounds proactive. You say “raising USD and serving global customers,” and it sounds like a serious business.
But often, there’s a simpler phrase behind “going global”: It’s too hard here; let’s see if there’s a way to survive out there.
That’s where Manus is most blindingly obvious. It wasn’t just going out to sell products, nor was it just earning a few dollars. It almost moved the entire capital narrative out: the entity is outside, the money is outside, the buyer is outside, and finally, even the exit was prepared to be completed outside.
This is no longer just “entrepreneurs working hard to find global markets.”
It’s more like telling those who come after: “Look, as long as you move fast enough and the story is smooth enough, you can sail the whole ship out.”
If this script works, the domestic VC circle will be excited, but some people might not be happy. After all, everyone pays lip service to encouraging innovation, going global, and building strong companies. But encouragement is one thing; it’s best to innovate in a place where everyone can see, manage, calculate taxes, and account for assets.
You can go fight overseas, but it’s best not to leave your provisions, your military honors, and your ledgers all overseas.
That is the subtlety of the Manus case.
It made many domestic AI entrepreneurs realize that the so-called “business environment” is never just about whether it’s convenient to register a company, whether office rent is expensive, whether there are policy subsidies, or whether local leaders come to take photos. The deeper issue is: when you actually make something valuable, can you really dispose of it according to business logic?
Can you sell it? To whom? How does the money come back? Whose technology is it? Where does the talent flow? Will this case be learned by others?
These questions aren’t usually put on the table. In meetings, everyone talks about supporting innovation, encouraging globalization, and firmly promoting high-level opening up—the atmosphere is very warm.
But when $2 billion, Meta, a Chinese team, a Singaporean entity, and AI technical assets all come together, the warmth cools down a bit.
So, Manus isn’t a “feel-good” story for the average person.
It’s more of a reminder for entrepreneurs: the domestic environment isn’t trying to stop you from making money, nor is it trying to stop you from going global, but you’d better understand that whose money you earn, whose technology you use, whose people you take, who you sell to, and where the value finally settles—these things are not decisions you can make behind closed doors as a lone entrepreneur.
If you’re just going out to sell goods, everyone welcomes you.
If you’re planning to sail the ship, the cargo, the crew, the routes, and the ledgers all away, then don’t blame the people on the shore for suddenly wanting to check your port paperwork.
That’s not a pleasant thing to hear.
But that is essentially what the Manus chicken was brought out to explain.
