Tencent, which has taken over Manus, is not another anxious Meta?

High-poaching talent, widespread investments, and increased spending — are Tencent and Meta following the exact same playbook?

Behind Tencent's move to acquire Manus lies the same anxiety over its AI transformation.

Recent reports indicate Tencent is leading a Chinese capital consortium that plans to purchase all equity of AI Agent firm Manus from Meta at an approximate $2 billion valuation (roughly 13.6 billion RMB).

Additionally, Tencent will hold the largest stake without taking controlling interest, allowing Manus to continue operating independently from its Singapore base.

This marks the latest capital development after Meta's earlier acquisition of Manus hit regulatory hurdles.

The most intriguing element of Tencent's planned Manus acquisition is Tencent itself: why is the company buying into Manus for a second time, at the same $2 billion price tag?

First, on pricing: it's not about paying more — it's about getting the best possible deal.

According to reports, the Tencent-led Chinese consortium's offer remains $2 billion, equivalent to 13.6 billion RMB (calculated at a 6.8 exchange rate), matching exactly the price Meta agreed to pay for Manus previously.

Manus' parent company, Butterfly Effect, was founded in 2022 and originally developed the popular AI browser extension Monica before launching Manus. Prior to Meta's planned acquisition, Butterfly Effect completed two funding rounds totaling over $10 million: the first round backed by ZhenFund, and the second including HSG, Tencent, ZhenFund, and Wang Huiwen. After these two rounds, Butterfly Effect reached a $100 million valuation.

It's important to clarify that whether this price is high or low should be judged from Meta's perspective.

For starters, Meta didn't raise the price at all.

When Meta first acquired Manus for $2 billion, the startup had an annual recurring revenue (ARR) of roughly $100 million. But within half a year, boosted by Meta's massive advertising channels, Manus' ARR skyrocketed to $400-500 million.

In other words, Meta poured enormous resources into Manus after acquisition, multiplying its revenue several times over — yet it's selling the company back to Tencent's consortium for the exact same $2 billion price tag.

Second, Meta isn't losing money either: while the transaction value remains unchanged, Meta has already spun Manus' core technologies out into its own in-house products.

This scenario is analogous to someone buying a new car, driving it for a year, then selling it to another buyer at the original new-car price. The original owner, the new buyer, and the dealership all seem to come out ahead — but someone has to take a loss. Who really is the losing party here?

Next, let's talk about equity structure.

Based on details from public reports, Tencent holds the largest single stake but does not exercise controlling rights. Looking at the participants in this acquisition — Tencent, Sequoia, ZhenFund — these are essentially Manus' earliest investors, who all profited handsomely when Manus was sold to Meta initially.

Theoretically, when Manus was acquired by Meta, ZhenFund, Sequoia and other early backers already fully exited their investments with substantial returns.

This buyback will inevitably force these investors to give back a portion of those previously realized gains.

Tencent's role in this deal is also highly meaningful. Earlier, after Meta's Manus acquisition fell through, there were reports that Manus planned to pursue an independent public listing on the Hong Kong Stock Exchange. By not taking control, Tencent retains only a financial investor position, preserving flexibility for Manus' future IPO path.

Furthermore, Tencent's existing AI business already shares significant overlap and similarities with Manus' products. Tencent's investments in its "Lobster" AI project this year, and its recently popular work assistant tool Work Buddy, are highly comparable to Manus' offerings.

If Tencent had chosen to fully acquire Manus and integrate it directly into its internal ecosystem, the strategic value of that move would likely have been limited.

Instead, since Manus primarily targets global overseas markets, maintaining its independent operations creates a mutually complementary dynamic with Tencent's existing business.

Finally, looking at Tencent itself: Tencent and Meta have followed remarkably similar AI development paths over recent years, and both have opted for aggressive acquisition strategies.

Placing this deal within a broader context, Tencent and Meta face strikingly comparable operational circumstances.

Both Tencent and Meta are social media-native tech giants with hundreds of millions of users on their platforms, yet both have lagged behind the curve somewhat in the current large language model wave.

Judging from Tencent's recent moves, the company has adopted the exact same strategy as Meta: filling gaps in its AI capabilities by poaching top talent.

The most famous example was Meta offering elite AI scientists compensation packages worth $200-300 million over four years, successfully poaching Pom Romen, former head of Apple's foundational model team, with a $200 million total compensation deal. Even OpenAI CEO Sam Altman publicly complained about Meta offering some employees signing bonuses as high as $100 million.

Tencent's recent talent strategy mirrors Meta's approach closely.

In December 2025, Tencent poached Yao Shunyu, a former OpenAI researcher, appointing him as its Chief AI Scientist reporting directly to Tencent president Martin Lau, while also putting him in charge of both the AI Infrastructure department and the Large Language Model department.

This restructuring signaled the failure of Tencent's previous large language model efforts led by its old AI Lab.

Then on March 20, Tencent issued an internal notice officially dissolving the AI Lab, reassigning its staff to the Large Language Model department and the Industry-University-Research Collaboration Center. Tencent vice president Jiang Jie no longer holds the position of AI Lab director.

AI is currently Tencent's most strategically urgent and anxiety-inducing business area. Previously, Tencent launched its Yuanbao AI assistant and invested in its "Lobster" AI project, and is about to launch Xiaowei, an AI assistant integrated into the WeChat ecosystem. Despite repeated rounds of heavy investment in Yuanbao, the product has delivered underwhelming results.

When DeepSeek burst onto the scene in February 2025, Tencent immediately integrated Yuanbao with DeepSeek-R1, and leveraged its full internal product ecosystem to promote the assistant. Within a month, Yuanbao's daily active users surged 20 times, hitting 41.64 million monthly active users in March to rank third in the industry, even briefly overtaking Doubao to top Apple's free app download chart at its peak.

However, Yuanbao's momentum quickly faded: starting in April that year, its monthly active users plummeted sharply for two consecutive months, dropping 44.8% month-over-month in May. By October, its daily active users had collapsed to just 5.6 million — nearly 10 times lower than Doubao's 54.1 million DAUs.

During the 2026 Spring Festival holiday, Tencent poured more resources into Yuanbao, giving the product a temporary seasonal rebound, but it still failed to challenge the market dominance of the two leading consumer-facing AI products.

Meta has also faced repeated setbacks in its AI business.

Meta once pinned high hopes on its open-source large language model Llama, but the 2025 launch of Llama 4 failed to meet expectations. The widening gap between Llama and top-tier industry models ultimately pushed Mark Zuckerberg to initiate a sweeping internal team restructuring.

Beyond foundational models themselves, Meta also experimented with consumer-facing subscription services for its Meta AI chatbot, and launched AI agent products targeted at advertising clients.

Meta's overall AI strategy has largely underperformed: by early July this year, reports surfaced that Meta planned to rent out its excess idle AI computing capacity to external customers.

Zuckerberg admitted openly during an internal all-hands meeting that AI agent development progress over the previous four months had not accelerated as the company had projected.

What's interesting is that Tencent stepping in to acquire Manus at this price point effectively validates a longstanding market hypothesis: Manus' valuation was largely tied to a narrow strategic "window of opportunity" it occupied within big tech ecosystems. After Meta's acquisition fell apart, many industry observers debated whether Manus had simply benefited from timing — and that once that window closed, the same asset would no longer command the same price.

Tencent acquiring Manus at nearly the exact same valuation Meta agreed to pay years ago is, in a sense, concrete proof of that hypothesis.

Meta spent two years, poured tens of billions of dollars, and poached dozens of top-tier AI scientists to build its AI ecosystem. Now Tencent is adopting the exact same playbook: poaching top talent with generous compensation packages, massively increasing investment, and placing widespread strategic bets, in a race to fill the last remaining gaps in its AI application layer.

Whether this path will ultimately lead Tencent to the same outcomes Meta has experienced remains an open question — one that even Tencent itself cannot answer right now.

This article originates from WeChat public account "Big V Business", written by Zhou Ming, and published with authorization from 36Kr.