Vitalik Buterin

The Blockchain Native Database by Sebastian Lorenz

Good morning everyone. Today Sebastian Lorenz unveiled AMP, a purpose‑built blockchain‑native database designed to treat blockchain data as a first‑class domain rather than an afterthought. By constructing the system from the ground up on modern data‑processing primitives, AMP aims to reconcile the contradictory demands of low‑latency, real‑time application interfaces and the inherent uncertainty of append‑only, re‑org‑prone ledgers. Lorenz highlighted the fundamental mismatch between traditional databases and blockchain workloads: conventional systems view consistency corrections as failure recovery, while blockchain applications must constantly handle block replacements, transaction disappearances, and chain reorganizations. Rather than cobbling together off‑the‑shelf components, AMP leverages Apache Data Fusion for SQL parsing and execution and the Arrow columnar format for in‑memory processing and gRPC transport, delivering both high‑throughput batch queries and sub‑second streaming semantics that are re‑org aware. The demo showcased concrete performance claims: AMP can decode millions of smart‑contract events per second, peaking at roughly four million events per second on an Uniswap swap benchmark, and maintains sub‑second latency at chainhead even on high‑throughput chains like Arbitrum and Solana. It exposes standard SQL, supports user‑defined functions for custom logic, and offers Rust and TypeScript streaming clients with exactly‑once guarantees and durable checkpointing, all packaged as a single binary that scales from local development to globally distributed deployments. For developers and enterprises, AMP promises to eliminate the heavy operational plumbing traditionally required to index and query blockchain state. By providing a declarative, SQL‑centric interface that integrates with existing tooling such as Foundry and Hard‑Hat, it accelerates data‑driven dApp development, enables real‑time analytics, and opens the door for broader ecosystem adoption, including LLM‑powered insights and cross‑chain data products.