Founded in 2016 by Alexandr WangScale Ai specializes in the structuring, annotation and data validation intended for the training of automatic learning models. Unlike companies that directly design the models or material infrastructure, Scale intervenes upstream, on a critical but often underestimated layer, namely the quality and organization of data.
A positioning focused on the reliability of the data
The company offers services covering the entire data processing cycle, from manual annotation to the optimization of datasets for specific uses. Its customers are civil and public players developing large -scale artificial intelligence systems. Among them are Microsoft,, OPENAIas well as the United States Department of Defenseespecially as part of the program Task Force Limadedicated to securing the military uses of AI.
SCALE is not limited to the generation of generalist datasets but also adapts its data games to specific contexts, such as military simulation, the training of legal or medical models, or even the validation of on -board systems.
A high -growing actor
According to data published by Bloomberg, Scale Ai has generated $ 870 million in revenues in 2024with a forecast at $ 2 billion for 2025. This progression is part of a growing demand for annotated data to support the rise in generative AI models and autonomous agents.
Capitalization side, the company has been valued $ 14 billion During its last financing round in 2024. At the beginning of 2025, a secondary operation mentioned in the specialized press suggests a valuation that can reach 25 billion dollars. None of the actors concerned has officially commented on this information.
An issue of sovereignty in AI
Data labeling is a key step in the development of artificial intelligence systems. If open source models are multiplying and calculation providers are diversifying, the constitution of specialized, reliable and representative data corpus remains a rare competence.
In this context, Scale AI occupies a strategic intermediate position, Neither developer of models nor infrastructure provider, but essential link in the learning chain. This function gives it growing visibility with manufacturers and public institutions involved in the development or deployment of AI -based technologies.
