Gimlet Labs has secured $300 million in a Series B funding round, bringing its total valuation to $3 billion. The round was led by Andreessen Horowitz, with participation from Arm Holdings, Samsung Ventures, and Microsoft’s M12 fund. Total external investment in the startup has now reached $392 million.
The company provides a platform that optimizes large language model inference by disaggregating models into smaller modules. These components are then deployed on chip architectures specifically suited to their hardware requirements. The system utilizes AI agents and a custom compiler to automate code adaptation and optimization for various processors.
Gimlet offers its software as a managed service or in a serverless edition. The startup reports having received billions of dollars in customer orders, with its client base including a major cloud provider and a top-tier AI laboratory. The new capital will be used to increase infrastructure capacity for the serverless platform by several hundred megawatts.
Beyond software, Gimlet plans to enter the custom hardware market. The company is currently developing an inference-optimized server designed to operate without a traditional motherboard, allowing for deployment outside of standard data center environments.