Modular builds a unified AI inference platform — from GPU kernels to cloud APIs — designed to run AI workloads across heterogeneous hardware including NVIDIA, AMD, Google TPU, AWS Trainium, Qualcomm, and Apple Silicon. Founded by Chris Lattner and Tim Davis, the company offers its MAX inference framework (open-source and enterprise) alongside a hosted cloud inference service, enabling teams to deploy 1,000+ open-source models or their own custom models with no vendor lock-in. Modular was acquired by Qualcomm in July 2026 and is headquartered in Silicon Valley with offices in San Francisco, Los Altos, Boston, and Edinburgh.