Nvidia, Meta back $1b round of data annotation firm Scale AI

Photo credit: Scale AI’s LinkedIn page

Developing an AI model from scratch means handling large amounts of unstructured data. This is crucial for any model’s success, but it is often resource intensive with little margin for error. As AI models advance and become more accessible, these so-called “frontier” data would grow even more in terms of quality, quantity, and complexity.

Scale AI, which calls itself an AI data foundry services firm, helps AI model developers like OpenAI, Toyota, and even the US Department of Defense to label their data. Its platform acts as a foundry for the end-to-end AI lifecycle, helping clients collect, curate, and annotate data.

The company, which is based in San Francisco, has secured US$1 billion in a series F funding round, bringing its valuation to nearly US$14 billion. Backers include Amazon, Meta, and Nvidia, as well as Y Combinator, Founders Fund, and Intel Capital.

Since its founding in 2016, Scale AI has worked with companies like Microsoft, Meta, and Anthropic. Its tech has been integrated into AI offerings across industries like ecommerce, retail, marketing, and legal.

“Our vision is one of data abundance, where we have the means of production to continue scaling frontier large language models many more orders of magnitude. We should not be data-constrained in getting to GPT-10,” said Alexandr Wang, CEO and founder of Scale AI.

In a LinkedIn post announcing the news, Wang said that Scale AI will focus on expanding its cache of frontier data “that will pave our road to artificial general intelligence.”

The series F funds will go toward improving Scale AI’s model analysis work with enterprise clients and the US Department of Defense. It also bolsters the firm’s role in the DEFCON 31 red-teaming event.

See also: What I learned about genAI from failing to scale my AI startup