companies

Scale AI

Data-labeling and AI infrastructure company that turns raw data into the training and evaluation pipelines frontier AI labs and governments rely on.

Scale was founded in 2016 by Alexandr Wang and Lucy Guo through Y Combinator, initially focused on labeling camera and lidar data for self-driving car companies. As large language models took over the industry's attention, Scale pivoted its human-in-the-loop workforce toward reinforcement learning from human feedback (RLHF) and evaluation data for frontier model developers, becoming one of the largest suppliers of training and benchmark data to labs building large language models. In June 2025, Meta invested roughly $14.3 billion for a 49% non-voting stake in Scale, and founder Alexandr Wang moved to Meta to lead its AI efforts; former Chief Strategy Officer Jason Droege became CEO. The deal raised questions across the industry about whether competing AI labs could keep sending Scale their most sensitive training data, and some customers have since diversified toward rivals such as Surge AI and Labelbox. Scale has also expanded into government and defense work through Donovan, a platform for U.S. defense and intelligence customers, and holds contracts with the Department of Defense's Chief Digital and Artificial Intelligence Office that have grown into the hundreds of millions of dollars.

Founded

2016

Headquarters

San Francisco, United States

Sector

data