Solana's consensus upgrade Alpenglow will launch on mainnet-beta on October 21, reducing block finality time to 150 milliseconds.
October 13 — According to official sources, Solana research and development company Anza has announced that Alpenglow will go live on Solana mainnet-beta on October 21 at approximately 22:50 (UTC+8). This upgrade, which took 18 months to develop, will complete its migration within a single slot. Block finality time will be reduced from several seconds to approximately 150 milliseconds. Alpenglow will replace the TowerBFT consensus protocol that Solana has used since its mainnet launch. The upgrade combines the latest consensus research with erasure-coded data dissemination technology, aiming to accelerate block confirmation speeds across the global network to as close to the physical limits as possible.
Alpenglow is expected to provide more efficient underlying infrastructure for global capital markets operating around the clock, supporting high-bandwidth trading and instant settlement. For institutional users, the protocol can reduce transaction settlement times from the usual 1 to 3 business days to approximately 150 milliseconds, while providing cryptographic guarantees and enabling payment processing to be completed directly on-chain, reducing intermediaries and reconciliation needs. For developers, faster confirmation speeds are expected to enhance the responsiveness of Web3 applications, reducing waiting indicators and retry logic, making on-chain application interactions feel closer to traditional internet applications.
Additionally, Alpenglow will provide AI agents with an on-chain transaction environment better suited for high-speed machine execution. Combined with payment tools such as x402, AI agents are expected to complete payments and other on-chain operations more efficiently without permission, reducing friction in the transaction process. The upgrade will further strengthen Solana's support for high-frequency trading, institutional financial services, Web3 applications, and the AI agent economy.