Meta is investing $14.3 billion (€12.4 billion) in AI company Scale and bringing its CEO, Alexandr Wang, on board to help develop advanced “superintelligence.”
The move highlights Meta CEO Mark Zuckerberg’s push to strengthen the company’s AI initiatives amid stiff competition from Google and OpenAI.
The strategic partnership values Scale AI at over $29 billion (€25 billion), according to the company.
Scale will remain independent, while the deal significantly expands its commercial partnership with Meta, which will acquire a 49% stake in the startup.
Alexandr Wang will join Meta with a small team but will stay on Scale’s board. Jason Droege, former chief strategy officer and former executive at Uber Eats and Axon, will serve as interim CEO.
Focus on ‘superintelligence’
Meta CEO Mark Zuckerberg is doubling down on “superintelligence,” a concept rival firms call artificial general intelligence (AGI). This marks another major pivot for Zuckerberg, who in 2021 focused on the metaverse, rebranding the company and investing heavily in virtual reality technology.
Following the AI boom sparked by ChatGPT in 2022, big tech continues to recruit talent from innovative AI startups without full acquisitions. Microsoft, for example, hired Inflection AI co-founder Mustafa Suleyman to lead its AI division.
Alexandr Wang co-founded Scale with Lucy Guo in 2016 while still a 19-year-old MIT student. Backed by Y Combinator—then led by Sam Altman, now OpenAI’s CEO—Wang later left MIT, echoing Zuckerberg’s path of leaving college to launch a groundbreaking tech venture.
Read More: GitHub CEO Explains Why Human Programmers Remain Essential
Servicing LLMs
Scale’s core business has been providing the human labor needed to train AI systems, from labeling images for self-driving cars to fine-tuning large language models (LLMs) like OpenAI’s ChatGPT, Google’s Gemini, and Meta’s Llama. Its clients include major automakers such as General Motors and Toyota, as well as leading AI developers, including Anthropic, OpenAI, Meta, and Microsoft.
By offering a more specialized service than Amazon Mechanical Turk, Scale helps AI teams improve training data and test model performance. Wang has also cultivated ties with the U.S. government, securing Pentagon contracts and collaborating with federal AI initiatives, with former Scale executive Michael Kratsios having led Trump’s science and technology office.
Meta’s acquisition signals a shift, though it remains unclear how Scale’s existing partnerships will be affected. Unlike many rivals, Meta has made its Llama system open-weight, allowing public access and modification of its core AI components.
Scepticism around LLMs
Meta reports that over a billion people use its AI products monthly, but it is widely seen as trailing competitors like OpenAI and Google in consumer adoption of large language models (LLMs). Its most advanced model, Llama 4 Behemoth, previewed in April as “one of the smartest LLMs in the world,” has yet to be released.
Meta’s chief AI scientist, Yann LeCun, winner of computer science’s top prize in 2019, has questioned the industry’s focus on LLMs. Speaking at French tech conferences, he noted that current models struggle with reasoning, planning, and understanding the physical world—skills essential to intelligent behavior. Instead, Meta aims to develop AI systems with human-level or even superhuman capabilities.
LeCun co-founded Meta’s AI research division with NYU professor Rob Fergus, who recently returned to lead the lab after a five-year absence. Fergus emphasized Meta’s long-term commitment to creating human-level AI experiences that could transform interactions with technology.
Frequently Asked Questions
What is Meta’s investment in Scale AI?
Meta is investing $14.3 billion (€12.4 billion) in Scale AI and acquiring a 49% stake.
Who is joining Meta from Scale AI?
Alexandr Wang, Scale’s co-founder and CEO, will join Meta with a small team while remaining on Scale’s board.
What is Meta’s focus on superintelligence?
Meta aims to develop human-level or superhuman AI, beyond the capabilities of current large language models (LLMs).
What does Scale AI do?
Scale provides human-labeled data to train and fine-tune AI systems, supporting self-driving cars and LLMs like Llama.
Who are Scale AI’s customers?
Clients include Anthropic, OpenAI, Meta, Microsoft, General Motors, and Toyota, among others.
What is Meta’s approach to LLMs?
Meta releases Llama as an open-weight AI model and focuses on long-term AI research toward human-level intelligence.
How does the Meta-Scale deal affect AI competition?
The partnership strengthens Meta’s AI capabilities as it competes with Google, OpenAI, and other innovators in superintelligence.
Conclusion
Meta’s $14.3 billion investment in Scale AI signals a bold push toward superintelligence, combining top-tier talent, advanced large language models, and long-term AI research. The partnership positions Meta to compete with industry leaders while exploring human-level and potentially superhuman AI capabilities.
