According to a Bloomberg report, Meta briefly considered purchasing Perplexity AI, currently valued at about $14 billion, before opting instead to make a headline-grabbing investment in Scale AI.
Perplexity AI has quickly risen in prominence since launching in late 2022. Powered by GPT‑4.1, Claude, Gemini, and its own in-house models, the platform now processes around 30 million search queries per day and is backed by investors such as Jeff Bezos and Nat Friedman.
Meta’s discussions with Perplexity reportedly involved preliminary acquisition talks centered on integrating its advanced LLM-powered search capabilities into Meta’s AI toolkit. However, the talks fizzled out without financial terms being finalized, and both parties decided not to proceed.
Following the Perplexity detour, Meta decisively turned to Scale AI, a data-labeling leader with which it struck a $14.8–14.3 billion deal for a 49% stake. The agreement, valued at roughly $29 billion post-deal, brings Scale’s cofounder and CEO Alexandr Wang into Meta’s “superintelligence” research initiative.
The structure, equity rather than outright purchase, enables Scale to maintain operational independence while giving Meta privileged access to labeled data for training large language models. The deal also ensures Meta secures future data needs through an advance‑payment arrangement.
Meta’s big move through Scale AI marks its most significant external investment in AI to date, following their $19 billion WhatsApp acquisition. This data-first strategy is viewed as a response to challenges in scaling internal AI models, including setbacks with the Behemoth and Llama series.
However, the deal prompted concern among Scale’s other clients, including Google, OpenAI, and xAI, with some pausing or terminating contracts citing worries over shared data access.