Liquid AI

Liquid AI's model is licensing efficient intelligence into other companies' products and stacks. Enterprises and device makers license the Liquid Foundation Models for deployment where transformer-scale models are impractical — on devices, at the edge, in constrained compute environments — and LEAP, the customization and deployment platform, is the recurring software layer through which customers build, adapt, and ship applications on LFMs. The December 2024 Series A welded a hardware alignment into the model: AMD Ventures led the $250M round, and Liquid co-optimizes its models for AMD GPUs, CPUs, and AI accelerators, making AMD both investor and channel — Liquid's models become a reason to deploy AMD silicon, and AMD's customers become Liquid's licensing pipeline. Apollo, the on-device private AI app for iOS and Android, functions as a public proof of the on-device thesis rather than a major revenue line. The economics aim at per-deployment license fees with platform subscriptions on top — recurring, high-margin, and heavily leveraged to a small number of OEM and enterprise design wins, which is both the attraction of the approach and its concentration risk.