Applied Compute in Talks for $3B Round Led by Elad Gil

Applied Compute is a one-year-old startup that helps companies run and customize open-source models with their own data, and The Information reports it is in talks to raise hundreds of millions of dollars in a round led by investor Elad Gil at roughly a $3 billion valuation.

The mark-up is what stands out. The company closed an $80 million round in April at $1.3 billion led by Kleiner Perkins, with Elad Gil, Lux Capital, Greenoaks, NEO and Hana Bicapital already on the cap table. As recently as last September it was in a $500 million-valuation conversation. Doubling the paper value in a matter of months is aggressive even by the standards of the current AI funding environment.

Two details make this more than another markup story. The first is what the company sells. Instead of competing with the frontier labs, the founders Rhythm Garg, Linden Li and Yash Patil, Stanford graduates who worked on reinforcement learning at OpenAI including the o1 model, package fine-tuning services on top of open-source model weights, with reported customers such as DoorDash using them for custom agent work. The second is who is writing the check. Elad Gil moving from being a name in the last round to leading this one is a different level of commitment.

A few reasons to hold this loosely. Only The Information has the terms, the round has not closed, and 'hundreds of millions' obscures both the actual raise and the dilution. Neither the report nor Applied Compute has confirmed revenue at the new price, and the last round was reportedly priced against $12.8 million in annualized revenue.

If the round prices where reported, it is another signal that enterprise AI money is flowing to teams selling open-source customization rather than another proprietary model, and Elad Gil's willingness to anchor rather than dot the cap table suggests that pattern still has room to run.