Scale AI, Inc. is a private San Francisco company founded in 2016 by Alexandr Wang and Lucy Guo. It supplies training data and evaluations to AI labs and builds AI applications for enterprises and governments. Bloomberg reported 2024 revenue of about $870 million. Meta owns 49% after a $14.3 billion investment in June 2025 that valued Scale above $29 billion, and Francis deSouza has been CEO since August 10, 2026.
What does Scale AI do?
Scale runs two businesses. The data business provides expert-written training data, human feedback, reinforcement learning environments, robotics data and evaluations to AI labs. The applications business builds generative AI systems on the Scale GenAI Platform and Donovan for companies such as Mayo Clinic, BP and Allianz and for governments including the U.S. Department of War and Qatar.
How does Scale AI make money?
Labs pay for data and evaluation projects, while enterprises and governments sign multi-year contracts for AI applications. Scale does not publish segment results, but it said its data business turned profitable in the second half of 2025 and applications revenue more than doubled in that period. In 2025 the Department of War awarded Scale two contracts totaling nearly $200 million.
Who owns Scale AI?
Meta holds 49% with no voting rights. Founders, employees and venture investors such as Accel, Founders Fund and Index Ventures own the rest. Alexandr Wang, who left in June 2025 to lead Meta Superintelligence Labs, remains chairman.
What changed after the Meta deal?
OpenAI and Google wound down work with Scale, and in July 2025 the company cut 200 full-time employees (about 14%) and 500 contractors. Interim CEO Jason Droege refocused on government and enterprise customers, and Scale reported its biggest revenue year in 2025 with more than $1 billion in new business. Former Illumina CEO Francis deSouza took over as CEO on August 10, 2026.
Who are Scale AI's competitors?
In training data: Surge AI, Mercor, Turing, Invisible Technologies, Labelbox and labs' in-house teams. In applications: Palantir, cloud providers and consulting firms.
