Scale

Scale is a San Francisco-based AI infrastructure company, founded in 2016 by Alexandr Wang, that provides the data labeling and evaluation work behind many production AI systems. Its Data Engine sources and annotates training data — for generative AI models, autonomous vehicles, robotics, and mapping — using a human-in-the-loop approach staffed through its Remotasks and Outlier platforms, with the company stating that around a quarter of its contributors hold advanced degrees. Scale says a large majority of leading generative AI model builders rely on this data pipeline. Beyond raw data work, Scale offers a GenAI Platform for deploying AI systems in enterprise settings with human oversight built in, Scale Donovan for applied AI use cases, and evaluation tools including private leaderboards that let AI companies benchmark model capabilities against each other. Customers span technology companies, automotive firms building self-driving systems, healthcare, and government and defense agencies through a unit sometimes called Scale Federal; reported clients include OpenAI, Meta, and the US Department of Defense. Meta has taken a 49% stake in the company as part of a $14.3 billion investment. Because so much of Scale's output depends on a large distributed annotation workforce, quality and turnaround can vary with the availability and consistency of that human labor pool.