Announced Thursday, the deal marks Meta’s most aggressive AI maneuver yet and comes as the company races to catch up to competitors like Google, OpenAI, and Microsoft in the increasingly high-stakes AI race.
Scale’s Role and Independence
While Scale will retain its status as an independent company, Meta will now own a 49% stake, significantly expanding their commercial partnership. Scale's post-deal valuation now stands at over $29 billion.
Despite leaving Scale to join Meta, Wang will remain on the board of directors. In his place, Jason Droege — formerly Scale’s Chief Strategy Officer and a veteran of Uber Eats and Axon — will serve as interim CEO.
Scale’s technology and services have become foundational for building AI, particularly large language models (LLMs) like ChatGPT, Gemini, and Meta’s own Llama. By helping annotate and evaluate datasets, Scale has carved out a unique niche servicing nearly every major AI developer, including Anthropic, OpenAI, and Microsoft. It remains unclear how the Meta partnership will impact these existing relationships.
Meta’s New AI Focus: Superintelligence
Meta’s investment signals a sharp pivot toward what CEO Mark Zuckerberg has called the pursuit of “superintelligence” — a long-term vision of developing AI that can match or surpass human cognitive abilities. The term echoes similar ambitions from other major players, particularly OpenAI’s focus on AGI.
While Meta has released its Llama language models under an open-source license, it has been widely viewed as trailing in LLM-driven consumer applications. Despite previewing Llama 4 Behemoth in April, billed as “one of the smartest LLMs in the world,” the model has yet to be released.
Wang’s addition to Meta's AI leadership team suggests the company is stepping up its game to change that narrative.
From MIT Dropout to AI Power Broker
At just 17, Wang co-founded Scale in 2016 with Lucy Guo, securing backing from Y Combinator, then led by OpenAI’s Sam Altman. He dropped out of MIT to focus on the company — a path reminiscent of Zuckerberg’s own decision to leave Harvard to build Facebook.
What began as a platform to support autonomous vehicles by labeling street imagery evolved into a cornerstone of the AI revolution. Scale’s annotation work today supports the training and evaluation of major LLMs and extends into U.S. government contracts, particularly with the Pentagon.
Meta’s Research Philosophy Still Diverges
Despite the aggressive investment, not all within Meta are sold on today’s mainstream approach to AI. Yann LeCun, Meta’s Chief AI Scientist and a Turing Award winner, has been openly skeptical of LLMs’ ability to reason, plan, or understand the physical world. Instead, LeCun has long advocated for a bottom-up approach to building human-level or superhuman AI based on deeper reasoning architectures.
Speaking at VivaTech in Paris this week, LeCun alluded to Meta’s evolving AI strategy, saying the company now has “a clearer vision” for achieving superintelligence, but avoided discussing the Scale deal directly.
Context Within Broader Industry Moves
Meta’s move mirrors recent talent acquisitions by other tech giants. Microsoft hired most of Inflection AI’s leadership, including co-founder Mustafa Suleyman, while Amazon brought on Adept AI’s CEO and acquired licenses to its technology. Google, too, has integrated startup talent from companies like Character.AI.
With the Club World Cup of AI innovation in full swing, Meta’s latest play with Scale and Wang could be its attempt to get back in contention for a title it once seemed poised to dominate.
The next step will be delivering — not just in ambition, but in AI systems that meet or exceed the promise of superintelligence.