Chroma is open-source search infrastructure for AI applications. It supports vector, full-text, regex and metadata search and is mainly used to store and query embeddings in applications built on large language models, for example for retrieval-augmented generation. Chroma is available as a Python and npm package and runs embedded in your application, as a standalone server or as a hosted serverless service, Chroma Cloud, which the vendor says is SOC 2 Type II compliant. Single-tenant deployments and bring-your-own-cloud in the customer's own VPC also exist.
The core software is licensed under the Apache License 2.0, so you can host Chroma entirely yourself. The company behind it is based in San Francisco, so US jurisdiction applies to the managed offering. Chroma Cloud runs in two regions: AWS US East and GCP Europe West 1 in Belgium. A database is created in one chosen region and stays there; according to the documentation, data is stored and processed entirely within that region. To limit dependence on a US vendor, self-hosting on European infrastructure is possible.