Meta has agreed to pay $14.8 billion for a 49% stake in Scale AI, a leading artificial intelligence data firm. This represents one of the social media giant's largest acquisitions since its 2014 WhatsApp deal.
As part of the agreement announced in June 2025, Scale AI's founder and CEO Alexandr Wang will leave his position to join Meta, where he will lead a new 'Superintelligence' initiative. Scale announced that Wang would quit to become an executive in charge of this new unit inside the tech giant. Jason Droege, Scale AI's chief strategy officer, has been promoted to CEO to replace Wang.
The move signals Zuckerberg's determination to catch up in the AI arms race. "As the pace of AI progress accelerates, developing superintelligence is coming into sight," Zuckerberg wrote in an internal memo viewed by Bloomberg. Zuckerberg has grown increasingly frustrated that rivals like OpenAI appear to be ahead in both underlying AI models and consumer-facing apps. Meta's release of its Llama 4 AI models in April was not well received by developers, further frustrating Zuckerberg. The company had planned to eventually release a bigger and more powerful "Behemoth" model, but that has yet to be made available due to concerns about its capabilities relative to competing models.
The acquisition has created significant ripples throughout the industry. Google, one of Scale AI's largest customers, is cutting ties with the data-labeling startup following Meta's investment. Google had planned to pay Scale AI about $200 million in 2025 for human-labeled training data to develop AI models like Gemini, its ChatGPT rival. Now, it is seeking alternative providers to avoid potential exposure of proprietary data to Meta. OpenAI, another major Scale client and Meta rival, has also begun winding down its work with Scale in the wake of the deal. OpenAI is phasing out its work with Scale AI and cutting ties with the data provider, with the company stating it was already winding down its work with Scale AI ahead of Meta's announcement.
The exodus shows that in the frenzied race to build the smartest AI, access to the best human-annotated data has become a critical, and now fiercely contested, strategic chokepoint. At a basic level, data labeling is how an AI model learns. In the early days, this meant drawing boxes around pictures of cats. But as models have grown more sophisticated, so has the data they need. It's no longer enough to just feed a large language model the entire internet; to make it truly smart—to get it to reason, to perform complex tasks, to not "hallucinate"—it needs to be taught by experts.
Scale AI's business trajectory reflects growing demand for high-quality data infrastructure in AI. After generating $870 million in revenue in 2024, the company expects that figure to more than double to $2 billion by the end of 2025. Its latest valuation of $29 billion underscores just how vital data infrastructure has become in the broader AI value chain. The stakes are enormous, with Meta planning to spend up to $65 billion on AI infrastructure this year alone, building massive data centers to train increasingly powerful models.