Summary
- Pinecone BYOC now supports on-demand capacity mode, allowing users to create indexes without pre-sizing each one.
- On-demand indexes share a capacity pool that supports millions of namespaces across AWS, Google Cloud, or Azure environments.
- Dynamic sharding routes new writes to shards with available room and scales capacity by adding shards without data reshuffling.
- High-query namespaces can be isolated to dedicated read nodes while maintaining the zero-access security model.
Why it matters
This feature eliminates upfront capacity planning and idle hardware costs for vector database deployments in regulated cloud environments.