The end of AI anxiety lies in the shopping cart.

Recently, the news about Meta's aggressive acquisitions has been all over the place. It spent $14.8 billion to acquire a 49% stake in Scale AI and offered an annual salary of $100 million to poach researchers from OpenAI. Seeing Meta acquire Scale AI, SSI, Perplexity, and Runway makes me think of how girls like to go shopping when they're unhappy, quickly placing five orders on Taobao.

What can we interpret from Meta's acquisition spree? Mark Zuckerberg is really anxious. American companies are really willing to spend money. And the exit paths for American startups are truly smooth.

Why is Meta so anxious? It's afraid of falling behind and losing the future. Regarding the opportunities in AI, if the M7 (the seven major technology companies in the US stock market) can keep up, a market value of $10 trillion will be just the starting point. Otherwise, they may experience a "Nokia moment". Facing the expectation of a $10 - trillion market value, spending tens of billions of dollars to build an AI team today doesn't seem expensive.

Every technological revolution is a redistribution of the market share. In the last mobile - internet revolution, American tech giants took 92% of the market share, while startups took 8%. No startup threatened the positions of Meta, Google, or Amazon. This was a victory for the tech giants.

I roughly calculated that in China, tech giants took 60% of the market share, and startups took 40%. Among the startups, three have a market value of over $100 billion - ByteDance, Meituan, and Pinduoduo, which broke the dominance of BAT. This was a victory for startups.

01

Let's first sort out the process of Meta's acquisition spree.

After the release of LLaMA4, it underperformed and received a lukewarm response. The core team members kept leaving. Mark Zuckerberg was really in a hurry and entered the "founder mode". He personally initiated acquisitions and recruitment to build a new AI team, claiming that "Meta must reshape its leadership in AI".

Recently, Meta spent $14.8 billion to acquire a 49% stake in Scale AI, valuing the company at $30 billion. This was an acquisition focused on talent. As a condition of the deal, 28 - year - old Chinese - American genius Alexandr Wang, the CEO of Scale AI, joined Meta to lead the new "Super Intelligence Laboratory" and build a new AI team to develop a model with 3 trillion parameters. Scale AI will remain independently operated.

Before that, Meta tried to acquire the AI search company Perplexity but failed. Perplexity was valued at $14 billion. Meta had discussions with it about the acquisition, but many terms couldn't be agreed upon, and Perplexity withdrew from the negotiation. Then Meta turned to Scale AI.

Meta also wanted to acquire Safe Super Intelligence (SSI), a company founded by former OpenAI Chief Scientist Ilya, which was valued at $32 billion in the angel round. After the acquisition was rejected, Meta poached its CEO Daniel, and it succeeded. CTO Ilya had to send an email saying that "he would serve as the CEO".

Meta also discussed the acquisition of the AI video company Runway, which was valued at over $3 billion. But the negotiation stalled before reaching the offer stage due to differences in the transaction terms.

In the talent competition, Meta poached 8 top AI researchers from OpenAI, 4 of whom are Chinese. Each was offered a transfer fee of $1 billion. What does this mean? Real Madrid paid $80 million to sign Ronaldo from Manchester United, while Meta paid $1 billion to sign Yu Jiahui from OpenAI. Now, top researchers are more expensive than top football stars! OpenAI publicly accused Meta of "breaking into our house and stealing things".

Mark Zuckerberg's acquisition spree is really like a girl's shopping spree. The end of AI anxiety is the shopping cart. One of the main motivations for people to shop is the fear of losing their youth and charm. Similarly, tech giants are afraid of falling behind in technology and losing their glory.

02

It's not just Meta. Other tech giants are also on an acquisition spree.

Google acquired Character.ai for $2.5 billion. Founder Noam returned to Google with the core technical team to participate in the research and development of the next - generation model. Microsoft bought out Inflection for $650 million. Founder Suleyman joined Microsoft with the core employees to build a new AI department. Amazon acquired Adept, and founder David joined with his team to establish the AGI Autonomy department.

Most of these people come from Google. They have only developed models and don't even have successful products. But if you have technical strength and have made a name for yourself, tech giants are willing to pay the price and recruit the team through acquisition. This creates a win - win situation. Startups are acquired after two or three years of operation, and VCs either make money or break even. The exit path is really smooth. If tech giants fall behind in AI, they will be anxious to some extent. This kind of acquisition is like buying time with money. And top talents often become the cure for anxiety.

It can be seen that the current M&A paradigm in AI is to buy the company along with the people, purchasing the founder's imagination and the ability to lead an AI team.

Meta bought the genius teenager Alexandr Wang, hoping that he would lead the Super Intelligence Laboratory to develop the next - generation model. Scale AI will develop independently outside the system. Google bought Noam, the core author of Transformer. The Character product will develop independently outside the system. These acquisitions are specially designed to avoid antitrust supervision.

In contrast, during the mobile - internet era, Meta had three key acquisitions: It acquired Instagram for $1 billion, WhatsApp for $19 billion, and Oculus VR for $2 billion. These three acquisitions were all about acquiring assets, not teams. After a transition period of three or four years, the founders left with substantial profits, and these products were further developed.

03

Currently, in the very early stage of the AI industry, there are almost no assets to buy. Top talents are the assets. AI startups have become a form of "advanced job - hunting". Either you become a king independently, which has a very low probability, or you build yourself into an advanced talent IP through entrepreneurship and sell yourself at a high price to the giants.

In China, such an M&A culture has not yet formed. Two or three years after the emergence of ChatGPT, there have been no high - priced talent acquisitions at least for now.

Why do many Chinese AI projects try to transform into overseas companies and raise funds from mainstream US VCs? Besides the better payment habits of overseas users, a more friendly overseas competitive environment, and more available funds overseas, there is another rarely mentioned reason - overseas M&A is easier. As long as you prove your value, tech giants are really willing to pay a high price.

The financing of Chinese AI projects has formed a standard paradigm. They first raise funds in China for two or three rounds. When the product becomes popular, it can gain the recognition of mainstream US or Singaporean VCs. Then they take the opportunity to raise overseas funds. The core team moves overseas, and the company transforms into a pure overseas company - with an overseas structure, overseas customers, and overseas shareholders. HeyGen, Manus, and Genspark have all followed this pattern. It is said that Lovart is also on the way. Many projects are trying this path.

Actually, the goal is overseas exit. An IPO is the best outcome, and being acquired is also a good choice.

The treatment of top AI talents in the US is like that of NBA players, while in China, it's like that of CBA players. It's normal for the best CBA players to want to develop in the NBA.

In terms of talent acquisition, there are far fewer cases in China than in the US, but the Chinese cases seem to be more story - worthy. US talent M&A is like a game of chess, with a clear goal and high - value moves. Chinese cases are more like a game of Go, with seemingly unimportant moves at a low price but ultimately determining the outcome.

Tencent's key acquisition was Zhang Xiaolong. In 2005, Zhang Xiaolong sold Foxmail to Tencent. He was in charge of QQ Mail in Guangzhou and later created WeChat. Alibaba's key acquisition was Jiang Fan. In 2013, Youmeng was acquired by Alibaba for $80 million. Jiang Fan started to lead the wireless transformation of Taobao on mobile.

ByteDance started acquiring talents when it was still a startup. For example, Zhang Nan previously founded Tubatu, Chen Lin founded a weather app, and Shen Zhenyu founded TuChong. They all joined ByteDance through M&A. A very important acquisition for ByteDance was Musical.ly. Musical.ly merged with Douyin to form TikTok, opening up the overseas market. The two founders, Yang Luyu and Zhu Jun, have been in charge of multiple businesses. Now, Zhu Jun is leading the AI product department Flow to explore a series of AI products like Doubao.

Why do the acquired founders become the pioneers of new businesses? The reasons behind this are worth pondering. Do entrepreneurs have a stronger ability to start from scratch? Do they have a more accurate intuition in user insight? Can they take more risks? I'm not sure.

But there is a simple reason - entrepreneurship is much more difficult than working for a company. Many startups don't achieve great success, but it's not easy to reach medium - or small - scale milestones, which also shows strength. Entrepreneurs are often like wild animals by nature. It's difficult for them to start working on business in a big company from the very beginning. So, the big companies have to buy them in.

Currently, we are only in the first act of AI development - the era of large models. Most of the M&A cases we can see are related to top algorithm talents. There will be a second and a third act in AI development. The protagonists in each act will be different. Top AI product managers and AI infrastructure talents will also appear, and there will be acquisitions targeting them.

Let's make a prediction. Next, let's look forward to the first case of a US tech giant acquiring a Chinese AI startup team, specifically a team that has transformed from a Chinese company to an overseas one.

May the strong always create their own opportunities. Wish you all good fortune.

This article is from the WeChat official account "AI Voyage". Author: Wu Bingjian. Republished by 36Kr with permission.