From Wuhan to Silicon Valley: How an AI Company Covered a Decade's Journey of Giants in 9 Months?

比推发布于2025-12-30更新于2025-12-30

文章摘要

Meta has acquired the AI startup Manus for billions of dollars, marking its third-largest acquisition to date. Founded by Xiao Hong, a graduate of Huazhong University of Science and Technology, Manus started in Wuhan and launched its product in March of this year. Within just nine months, the company grew its annualized revenue to $125 million with a global subscription-based user base. Initially criticized as a "shell" for using third-party AI models like Claude and Qwen, Manus focused on the application layer rather than developing its own models. The acquisition process took only 10 days, as Meta urgently sought a proven AI product to compete with OpenAI’s ChatGPT and Google’s Gemini. A key turning point was Manus’s relocation from China to Singapore in July, which involved downsizing from 120 to 40 employees. This move facilitated the acquisition amid geopolitical complexities. Investors, including Sequoia China, Tencent, and ZhenFund, saw returns of dozens of times their initial investments. Xiao Hong will now serve as a VP at Meta, reporting directly to Mark Zuckerberg. The deal underscores a new paradigm: Chinese entrepreneurs and investors can achieve global success, though it may require strategic relocations to navigate international regulations.

Author: Wawa, Deep Tide TechFlow

Original Title: From Wuhan to Silicon Valley, Manus Took Nine Months


Today's biggest news in the AI circle: Meta acquires Manus for billions of dollars.

This is Meta's third-largest acquisition in history, second only to WhatsApp and Scale AI, and even more expensive than the acquisition of Instagram back in the day.

Looking at Manus's timeline, the product was only launched in March this year, and it was acquired in December. From release to sale, it took just 9 months.

Founder Xiao Hong, from Ji'an, Jiangxi, graduated from Huazhong University of Science and Technology and started his business in Wuhan. His first product was the WeChat public account formatting tool Yiban, which was sold. The second product was the enterprise WeChat CRM tool Weiban, which was also sold. The third product was the browser AI plugin Monica, which wasn't sold but was criticized.

Criticized for what? For being a shell.

At that time, the industry consensus was that only companies developing large models had a future, and those building applications on top of others' models were just shells with no technical substance.

When Manus first became popular in March this year, co-founder Ji Yichao responded to a netizen's question on social media: "We have used Claude and also fine-tuned different versions of Qwen (Qianwen) models."

In other words, they used others' large models and built the application layer themselves.

So what?

Now it's worth billions of dollars.

Last year, ByteDance executives flew to Hong Kong specifically to meet Xiao Hong, offering $30 million to acquire the company. Xiao Hong didn't sell.

Looking back now, the difference between $30 million and billions is not just a year's time; it's:

A product that was delivered.

Additionally, we think the most interesting part of this story is not the ending but the process.

In July this year, Manus made a decision: to move the company from China to Singapore. Out of a 120-person team, only 40 core technical staff were retained to move together, and the rest were laid off. The Beijing office was closed, and the Wuhan office was also closed.

At the time, many people criticized them for "running away."

Now, it seems this step was necessary. It is almost impossible for a Chinese company to be acquired by a U.S. tech giant under current regulatory conditions. Changing the registration location removed the obstacle.

The negotiation took only 10 days.

ZhenFund partner Liu Yuan said it was so fast that they initially doubted whether it was a fake offer.

How desperate was Meta to close a multi-billion-dollar deal in just 10 days?

Looking at the background: This year, Meta's capital expenditure on AI exceeded $70 billion, but most of it was spent on infrastructure, with few products ready for use. OpenAI has ChatGPT, Google has Gemini, what does Meta have?

Llama is open-source, and anyone can use it. Meta needed a competitive application-layer product, and Manus happened to be ready-made.

Annualized revenue of $125 million, built from scratch in 8 months, global users, subscription-based, and proven viable.

This is not just acquiring a team; it's acquiring a:

Validated business model.

Interestingly, Manus's investor list includes Sequoia China, Tencent, and ZhenFund. When these investments were made, the valuation was in the tens of millions of dollars. Now, upon exit, the returns are dozens of times higher.

So, you see, Chinese VCs invested in a Chinese company, the company moved to Singapore, was acquired by a U.S. company, and Chinese VCs made money from the U.S. company.

This chain is even more Agent than Manus's product.

After the acquisition, Xiao Hong will become a vice president at Meta. A entrepreneur who started in Wuhan with a WeChat public account formatting tool is now going to report to Zuckerberg in Silicon Valley.

ZhenFund's Liu Yuan said: "The era for this generation of young Chinese entrepreneurs has arrived."

This statement might only be half right.

The era has indeed arrived, but the way it arrived was by moving the company away.


Twitter:https://twitter.com/BitpushNewsCN

BitPush TG Discussion Group:https://t.me/BitPushCommunity

BitPush TG Subscription: https://t.me/bitpush

Original link:https://www.bitpush.news/articles/7599188

相关问答

QWhat is the name of the AI company acquired by Meta and for how much?

AThe AI company acquired by Meta is called Manus, and the acquisition was for several billion dollars, making it Meta's third-largest acquisition in history.

QHow long did it take from Manus product launch to its acquisition by Meta?

AIt took 9 months from Manus' product launch in March to its acquisition in December.

QWho is the founder of Manus and what was their background?

AThe founder of Manus is Xiao Hong, who is from Jiangxi, China, graduated from Huazhong University of Science and Technology, and previously created products like a WeChat public account formatting tool and an enterprise WeChat CRM tool.

QWhy did Manus move its company from China to Singapore before the acquisition?

AManus moved its company from China to Singapore to avoid regulatory hurdles and make the acquisition by a U.S. tech giant like Meta feasible, as it would be nearly impossible for a Chinese company to get approval for such a deal under current geopolitical conditions.

QWhat was the business model of Manus that made it attractive to Meta?

AManus had a subscription-based business model with an annualized revenue of $125 million, global users, and a proven product-market fit, which provided Meta with a ready-to-use application layer product in the AI space.

你可能也喜欢

Solana 基金会与 Google 合作推出 Pay.sh,能否打通智能体经济中 Web2 与 Web3 的支付链路?

随着AI智能体能力的增强,面向智能体构建支付系统成为必要,但现有方案存在局限:传统支付系统(如信用卡)不适合智能体;新兴链上支付协议(如x402、MPP)则需额外搭建系统,门槛较高。两者未能融合,限制了智能体服务的范围。 为此,Solana基金会与Google Cloud联合推出Pay.sh,定位为“智能体与企业级服务之间的支付网关”,旨在打通智能体调用服务的支付环节。用户可通过信用卡或稳定币为Solana钱包充值,该钱包即可作为智能体的身份和支付代理。智能体无需注册或输入API密钥,即可通过Pay.sh使用Google Cloud、阿里云等服务。 Pay.sh兼容x402和MPP协议:当服务器返回402状态码(需要支付)时,Pay.sh能根据服务类型(一次性付费或持续计费)自动选择合适的支付方式。它支持一次性转账或会话授权凭证,以降低使用成本。同时,Pay.sh确保钱包本地存储,仅在支付时请求用户确认,并过滤服务商返回的指令以防止恶意攻击。 对服务商而言,Pay.sh提供易于集成的网关,只需声明支付参数即可支持复杂场景(如阶梯收费、自动分账)。服务商还可将API发布至Pay Skill Registry,供智能体发现和使用。 Pay.sh的优势在于打通Web2与Web3支付生态,为智能体提供合规身份和支付能力。通过与Google合作,其流量调度和日志合规可在Google Cloud上完成,有助于规范智能体行为。此外,Pay.sh补全了智能体商业协议(如A2A、AP2)的支付环节,促进了价值流动,也为Solana生态带来更多流量。 但Pay.sh仍面临挑战:服务商注册表缺乏准入和去中心化验证机制,可能导致智能体接入冒牌服务;其安全性依赖底层协议,存在外部风险;同时,各国API供应商可能因合规顾虑而限制使用。 总体而言,Pay.sh标志着智能体支付基础设施向Web2与Web3融合迈出关键一步,链上钱包有望成为智能体参与多样化任务的背书,其后续发展值得关注。

marsbit14分钟前

Solana 基金会与 Google 合作推出 Pay.sh,能否打通智能体经济中 Web2 与 Web3 的支付链路?

marsbit14分钟前

如何用Claude Skills自动化任何工作流(完整教程)

本文是一份全面的Claude Skills实战教程,旨在帮助用户通过构建自定义Skills(技能),将AI深度、稳定地融入日常工作流,实现任务的标准化与自动化。 **核心概念:** Claude Skill并非简单的预设提示词,而是一个包含详细工作流指令的永久性文件(SKILL.md)。它像一个训练有素的“员工”,能确保每次执行相同任务时,流程、标准和输出格式都保持一致,从而获得远超一次性Prompt的稳定高质量结果。 **教程分为四个阶段:** 1. **快速上手(5分钟):** 了解Skills以文件夹形式存放,学会从官方或社区平台查找、安装现有Skill,并应用于真实任务进行初步体验。 2. **从零构建:** 在创建自定义Skill前,需明确回答三个问题:技能的具体用途、至少五个触发短语、完美输出的具体示例。随后编写SKILL.md文件,其结构包括:顶部的YAML元数据(含名称和具体触发描述)和下方用自然语言编写的一步步工作流说明(需包含示例、边界情况及明确标准)。 3. **测试与优化:** 通过“三场景测试”确保Skill达到生产级水平:常规场景(80%常用情况)、边界场景(处理异常或信息缺失)、压力场景(应对最复杂情况)。每次不理想的输出都应触发对SKILL.md的即时优化,通过持续迭代使输出媲美专业人士。 4. **搭建技能库:** 为工作流中各项重复任务构建专属Skill,例如内容创作、数据分析、邮件起草等。一个月可积累十个生产级Skill,三个月可形成覆盖主要工作的完整技能库,从而将AI转化为高效的“团队”,让使用者更专注于策略而非执行。 **文章强调,** 持续构建和优化Skill库能带来显著的时间复利(例如十个技能每年可节省260小时),是从“使用AI”转向“用AI系统工作”的关键。

marsbit42分钟前

如何用Claude Skills自动化任何工作流(完整教程)

marsbit42分钟前

交易

现货
合约

热门文章

如何购买S

欢迎来到HTX.com!我们已经让购买Sonic(S)变得简单而便捷。跟随我们的逐步指南,放心开始您的加密货币之旅。第一步:创建您的HTX账户使用您的电子邮件、手机号码注册一个免费账户在HTX上。体验无忧的注册过程并解锁所有平台功能。立即注册第二步:前往买币页面,选择您的支付方式信用卡/借记卡购买:使用您的Visa或Mastercard即时购买Sonic(S)。余额购买:使用您HTX账户余额中的资金进行无缝交易。第三方购买:探索诸如Google Pay或Apple Pay等流行支付方法以增加便利性。C2C购买:在HTX平台上直接与其他用户交易。HTX场外交易台(OTC)购买:为大量交易者提供个性化服务和竞争性汇率。第三步:存储您的Sonic(S)购买完您的Sonic(S)后,将其存储在您的HTX账户钱包中。您也可以通过区块链转账将其发送到其他地方或者用于交易其他加密货币。第四步:交易Sonic(S)在HTX的现货市场轻松交易Sonic(S)。访问您的账户,选择您的交易对,执行您的交易,并实时监控。HTX为初学者和经验丰富的交易者提供了友好的用户体验。

2.2k人学过发布于 2025.01.15更新于 2025.03.21

如何购买S

相关讨论

欢迎来到HTX社区。在这里,您可以了解最新的平台发展动态并获得专业的市场意见。以下是用户对S(S)币价的意见。

活动图片