Applied Compute Raises $350 Million at $3.25 Billion Valuation in Round Led by Elad Gil

Angel investor Elad Gil led the round, according to the report, with Nvidia, Shopify Ventures, HubSpot Ventures and more than a dozen other investors taking part.

Applied Compute said its annualised recurring revenue has passed $100 million. The company launched less than 18 months ago and counts Microsoft and DoorDash among its customers. It plans to use the new money to expand its market presence and buy more infrastructure.

What the Platform Does

The company runs a cloud platform on which developers train LLMs and serve them for inference, supported by an in-house team of machine learning researchers who help customers with work such as building quality evaluations for newly trained models.

Its main technical pitch is asynchronous reinforcement learning. In a typical training setup, one component runs many copies of a model on different tasks to generate training data, while a second component uses that data to adjust the model's weights, and each waits for the other to finish. Applied Compute's system lets both run at the same time, which the company says speeds up development. The platform also supports self-distillation, a method that removes the need for a separate teacher model to supervise training.

For inference, an autoscaling engine adjusts the hardware assigned to a model according to user traffic, and an online reinforcement learning feature periodically fine-tunes a deployed model using the prompts it receives. The company also offers Shift-1, a model router that sends each coding agent request to the model that can handle it most cost-efficiently and can split complex tasks across several models.

"Training, inference, and routing are the first pieces of a much larger stack," founders Rhythm Garg, Yash Patil and Linden Li wrote in a blog post, adding that they are building infrastructure for companies to "fully own their intelligence."

Why It Matters

The round is another large bet on the layer of companies that sit between chipmakers and businesses using AI. It comes in the same week that inference provider Fal said it had doubled its valuation to $16 billion in a round that Nvidia also joined. Applied Compute's pitch rests on the idea that companies will want to train and continuously update their own models rather than rely only on frontier labs' APIs, and its revenue figure suggests some large customers are already paying for that approach.