From Wuhan to Silicon Valley, Manus did it in just nine months.

Source: TechFlow (Shenchao)

Today’s biggest news in the AI world: Meta acquires Manus for several billion dollars.

This is Meta’s third-largest acquisition in its history, behind only WhatsApp and Scale AI—and more expensive than its acquisition of Instagram.

Looking at Manus’s timeline, the speed is striking. The product launched in March this year and was acquired in December. From release to exit: just nine months.

The founder, Xiao Hong, is from Ji’an, Jiangxi, and a graduate of Huazhong University of Science and Technology. His entrepreneurial journey began in Wuhan.

His first product was Yiban, a WeChat content formatting tool—sold.

The second was Weiban, a WeCom CRM product—also sold.

The third was Monica, a browser-based AI plugin—not sold, but heavily criticized.

Criticized for what?

For being a “wrapper.”

At the time, the prevailing industry belief was that only companies building large models had real prospects. Applications built on top of others’ models were dismissed as “shells” with no technical depth.

When Manus first went viral in March, co-founder Ji Yichao responded to questions on social media, saying:

“We’ve used Claude, and we’ve also used different fine-tuned versions of Qwen.”

In other words, they used other companies’ foundation models and focused on the application layer.

So what?

That approach is now worth billions of dollars.

Last year, senior executives from ByteDance flew to Hong Kong to meet Xiao Hong, offering USD 30 million to acquire Manus. He declined.

In hindsight, the gap between USD 30 million and several billion is not one year of time. It is this:

A product was truly built.

What makes this story most interesting is not the ending, but the process.

In July this year, Manus made a decisive move: relocating the company from China to Singapore. Out of a 120-person team, only 40 core technical staff moved together; the rest were laid off. The Beijing office was shut down, and the Wuhan office was closed as well.

At the time, many criticized the company for “leaving.”

Looking back, the move was necessary. Under current conditions, it would be nearly impossible for a Chinese company to be acquired by a U.S. tech giant and pass regulatory approval. Changing the place of incorporation removed that barrier.

The negotiation took just 10 days.

Liu Yuan, a partner at ZhenFund, said it was so fast that they initially suspected the offer might be fake.

Closing a multi-billion-dollar deal in 10 days—how urgent was Meta?

The context is revealing. This year, Meta’s capital expenditure on AI exceeded USD 70 billion, but most of it went into infrastructure. In terms of usable, consumer-facing products, there were few standouts.

OpenAI has ChatGPT.

Google has Gemini.

What does Meta have?

Llama is open source—anyone can use it. Meta needed a strong application-layer product, and Manus was ready-made.

Annualized revenue of USD 125 million, built from zero in eight months, global users, subscription-based, fully validated.

This was not an acquisition of a team. It was an acquisition of a proven business model.

Another interesting detail: Manus’s investor list includes Sequoia China, Tencent, and ZhenFund. When they invested, the valuation was in the tens of millions. At exit, returns were dozens of times that.

So the chain looks like this:

Chinese VCs invest in a Chinese company → the company relocates to Singapore → it is acquired by a U.S. company → Chinese VCs earn money from an American acquirer.

That chain is more “agent-like” than Manus’s own product.

After the acquisition, Xiao Hong becomes a Vice President at Meta. A founder who started in Wuhan, building a WeChat formatting tool, is now heading to Silicon Valley to report directly to Mark Zuckerberg.

Liu Yuan of ZhenFund summed it up with one line:

“An era for China’s new generation of young entrepreneurs has arrived.”

That statement may only be half true.

The era has indeed arrived—but it arrived by moving the company elsewhere.