Baseten

Baseten is an AI inference platform founded in 2019 by Tuhin Srivastava, Amir Haghighat, Phil Howes, and Pankaj Gupta after the founders saw strong machine learning models repeatedly get stuck in "deployment hell." The platform gives companies the infrastructure and engineering depth to train, serve, scale, and orchestrate hardware for AI systems in production, functioning much like a managed compute layer purpose-built for running open-source, fine-tuned, and custom models as production-ready APIs.

Baseten has grown rapidly alongside enterprise demand for AI inference: by mid-2026 the company said its platform was processing more than one billion inference calls per day across dozens of GPU clusters spanning multiple clouds. It has struck cloud and compute partnerships with AWS, Vultr, and NVIDIA, and vertical partnerships such as one with biotech software company Benchling to bring scalable GPU inference to scientific R&D customers.

Headquartered in San Francisco, Baseten has raised capital across six rounds in a little over four years, with its valuation climbing from roughly $200 million in 2024 to $13 billion by June 2026 as investor interest in AI inference infrastructure accelerated.