Neon's implementation of storage sharding for Postgres aims to enhance scalability by separating storage and compute, with compute handled by virtual machines and pods that autoscale on-demand. The storage component uses a Rust-based, multi-tenant engine called Pageserver to store historical data, with sharding distributing data across multiple Pageservers to increase capacity and throughput. This approach prevents large and small databases from competing for resources on the same server and offers linear scalability of storage I/O with dataset size. Storage sharding is especially beneficial for applications requiring high storage capacity and read-intensive workloads. Neon's architecture allows for the storage of tens to potentially hundreds of terabytes, with plans for future optimizations like shard splitting and enhancing shard subscription to Safekeepers. This system is poised to support even larger datasets and higher throughputs, marking a significant step toward achieving bottomless storage.