Applied Compute Raises $350M at $3.25B Valuation Led by Elad Gil

What the platform does

Applied Compute was founded by Yash Patil, Rhythm Garg and Linden Li. It offers an AI cloud that lets companies train and run large language models (LLMs). Developers can manage both training and inference on a single infrastructure, and the company also supports customers with machine learning research.

One of the platform's main features is asynchronous reinforcement learning, which speeds up training by running data generation and model weight updates at the same time. The company also uses self-distillation, with which it aims to remove the need for teacher models during training.

Beyond training and inference, Applied Compute offers a model router called Shift-1. The tool sends requests from coding agents to the most cost-effective AI model and can split complex tasks across several models. The platform automatically scales hardware resources based on user traffic and supports regular fine-tuning so models keep adapting to customer needs.

Fast growth

The new round follows an $80 million investment the company raised in October 2025. The jump in funding and valuation in about a year reflects continued investor appetite for infrastructure that lets businesses build and operate their own models rather than rely only on general-purpose ones.