Modal Labs, Inc. is an American cloud company based in New York. Available at modal.com, its platform offers a serverless computing environment: developers describe their workloads and resource requirements, while the service handles execution and scaling. The company focuses particularly on artificial intelligence use cases, where GPU access and software environment management present significant challenges.
Roots in developer tools
Founded in 2021 by Erik Bernhardsson and Akshat Baid, Modal Labs starts from an observation: bridging the gap between a program that runs on a computer and a production-ready service can require considerable infrastructure work. Erik Bernhardsson worked at Spotify and Better.com, among other companies, before founding the business. This experience informs an approach centered on technical team productivity.
Modal thus seeks to bring development and operations closer together. Rather than asking users to configure machines or a Kubernetes cluster themselves, the platform allows them to define much of their environment directly in Python. This approach aims to reduce the steps needed to move from experimentation to recurring workloads or remotely accessible services.
From serverless computing to artificial intelligence applications
The core offering involves running functions in containers, with CPU resources, memory and, if necessary, GPUs. Developers can specify software dependencies, use persistent storage and manage the secrets required by their applications. Modal also provides mechanisms for running scheduled tasks, parallelizing workloads and exposing HTTP endpoints.
These capabilities cover several use cases: data preparation, batch computing, running machine learning models and inference for generative applications. The infrastructure’s elasticity allows resources to be adjusted to demand, rather than keeping dedicated servers running at all times. Billing is based in particular on the resources consumed, under the terms of the chosen plan.
The promise, however, is not to eliminate all engineering work. Teams remain responsible for their models, code, data and performance monitoring. Choosing such a platform also requires examining latency, actual costs at scale and dependence on the provider’s interfaces.
What next?
Modal operates in a competitive market, between major general-purpose cloud providers and suppliers specializing in AI computing. Its growth will depend on its ability to combine ease of use, resource availability and cost control. For customer companies, decision criteria will extend beyond GPU access alone: operational reliability, security, data location and integration with existing tools will also carry weight. The challenge will be to turn ease of prototyping into a lasting foundation for production applications.