What Meta Gains From Scale CEO Alex Wang

The cost of acqui-hiring talent keeps rising! Meta Platforms agreed to shell out close to $15 billion for a 49% stake in data-labeling startup Scale AI, whose CEO Alexandr Wang will be joining Meta in a senior position, my colleague Cory Weinberg reported yesterday. 

I spoke with a number of Scale investors yesterday who shared their uncertainty about what the deal means for the company’s future.

Just as important, though, are the implications for the broader industry that Scale has been leading. 

AI developers such as OpenAI, Anthropic, Google and Meta have paid startups like Scale billions of dollars in recent years to hire countless people to teach models how to improve in a variety of domains, from generating good code to writing good prose. The human raters work on these models after they are trained on data from the web and other sources—hence, Scale and its ilk are known as post-training firms.

It’s not clear what Scale’s customers, including Microsoft, Amazon, Nvidia and OpenAI will do in the wake of the Meta deal. Some might be concerned that Scale could give away their secret post-training strategies to rival model developer Meta. If so, this deal could be good news for rivals such as Turing and Invisible in the short term.