Cloud infrastructure for AI agents
Morph Labs
Morph Labs builds Morph Cloud, infrastructure for running and developing AI agent applications. Its Morph Devboxes, powered by Infinibranch, give developers programmable, shareable virtual machines for coding agents, remote desktops, and web applications. Developers can manage environments through an API, interact with persistent agent sessions through SSH or a browser, and connect local editors. Devboxes can branch into hundreds of parallel replicas or scale to zero while preserving application and memory state.
Founded by Jesse Michael Han, previously a researcher at OpenAI, Morph Labs worked on models that generate code in the Lean theorem-proving language. Its infrastructure also supports formal mathematics: Morph partnered with Math Inc to develop Trinity environments for Math Inc’s Gauss autoformalization agent. Math Inc credits Infinibranch on Morph Cloud with supporting a workload involving thousands of concurrent agents, each with its own Lean runtime, consuming multiple terabytes of cluster memory.
Morph sells access through free and paid subscription tiers with metered resource credits and additional pay-as-you-go usage. Devboxes and GitHub Actions runners draw from the same compute-unit pool. Teams can connect environments to local tooling, remote agents, or CI pipelines; full virtual machines support container stacks such as local Kubernetes clusters and self-hosted Supabase.
1 talk
Newest first1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Messages from the stage
Branching and scaling agent workloads
Han introduced Morph Liquid Metal with faster branching, container support, and elastic autoscaling. GPU support was described as planned.
Affiliations reflect each recorded session, not necessarily current employment.
