Helping Enterprises Train Their Own AI, Applied Compute Valuation Surges to $3 Billion

According to Dongcha Beating's monitoring, The Information cited insiders as saying that Applied Compute is in talks for a new round of financing with a valuation of around $3 billion. The expected funding size is hundreds of millions of dollars, with investor Elad Gil in talks to lead the investment. Founded by three former OpenAI researchers, the company mainly helps enterprises train, deploy, and continuously improve open weight models with their own data.

Just four months ago, Applied Compute had a post-money valuation of $1.3 billion and completed an $80 million financing round led by KP21. The company's current annual revenue is around $50 million, nearly quadrupling from the $12.8 million disclosed by CEO Yash Patil in November last year.

Applied Compute offers a full suite of services to "train ready-made models into enterprise-specific models." DoorDash used it to train a menu correction model, reducing the proportion of low-quality menus in A/B testing by about 30%, which has since been deployed to all menu traffic across the United States.

Enterprises are also starting to turn to customizable and deployable open models to reduce the cost of using OpenAI and Anthropic. Applied Compute has perfectly tapped into this surge in demand.