Modal is a New York-based serverless compute platform that specializes in high-performance AI infrastructure. It enables developers and data teams to efficiently run compute-intensive workloads such as machine learning inference, training, and batch processing without the need to manage underlying infrastructure. Founded by Erik Bernhardsson and Akshat Bubna, Modal has expanded its presence with offices in New York, Stockholm, and San Francisco, achieving unicorn status with a valuation of $1.1 billion in September 2025.
The platform allows users to turn Python functions into autoscaling cloud workloads with GPU acceleration. Key features include a custom Rust-based container runtime, dynamic scaling across CPUs and GPUs, and real-time log streaming. Modal offers a comprehensive suite of services that covers the entire machine learning lifecycle, including model deployment, training, batch processing, and collaborative workspaces. It primarily serves AI/ML-focused technology companies, scientific research, and media sectors, with a pay-per-use pricing model that facilitates easy adoption for machine learning engineers and data teams.