The change is to finality, the point at which a transaction can no longer be reversed and the moment an exchange, bank or merchant waits for before treating an asset as moved. Under TowerBFT, Solana gave a conditional confirmation within seconds but took about 12.8 seconds to reach full finality. Alpenglow condenses those two steps into one.
“This is the largest, most ambitious blockchain upgrade since ‘The Merge’ and Anza took it from a white paper to running on mainnet in under 18 months,” said Anza CEO Brennan Watt, referring to Ethereum’s 2022 switch to proof of stake.
Before reaching mainnet, Alpenglow went through public testing, including red teaming, penetration testing, repeated migration rehearsals and a security competition, followed by launches on Solana’s testnet and devnet. Anza implemented the protocol in Agave, the validator software it maintains, and independent validators activated it.
Anza’s research team, led by ETH Zurich professor Roger Wattenhofer with Quentin Kniep and Kobi Sliwinski, published the Alpenglow design in May 2025.
Under Alpenglow, validators send their votes directly to each other instead of writing them into later blocks, and votes are weighted by stake. “If you have 80% of the votes say yes, then it’s immediately done,” Wattenhofer, Anza’s head of research, said in an interview with Solana Foundation technology vice president Jacob Creech published last month. “You immediately know that the block is final.”
A second route, two rounds of voting that each need 60% of stake, can be even quicker for validators clustered close together, such as in Europe, he explained. The protocol is designed to keep working with less than 20% of stake acting maliciously and a further 20% offline. If the network goes 10 seconds without finalizing a block, validators gradually lengthen block times until messages get through. “Eventually, even a bad internet will give us a block again,” Wattenhofer said.
More upgrades are planned. Rotor will change how block data spreads across the network, and geographically aware leader rotation aims to cut delays as block production passes between validators. Quantumglow, a research design, would swap Alpenglow’s elliptic-curve cryptography for hash-based signatures to make it secure against quantum computers while keeping the same finality speed. None has a committed date.
Alpenglow’s road to mainnet
Validators approved the upgrade in September 2025, voting 98% in favor of SIMD-0236, the proposal that implements it. “The biggest difference users will feel right away will be a reduction in confirmation latency to approximately 150-200ms,” Anza lead economist Max Resnick told Decrypt at the time.
Alpenglow went live on Solana’s community test cluster in May 2026, when Resnick said the team was targeting a mainnet activation in late Q3 or early Q4. On October 13, Anza confirmed the October 21 date.
The upgrade also opens the door to other changes. Developers at Jump Crypto’s Firedancer team have proposed scrapping Solana’s cap of 60 million compute units per block once Alpenglow is live, letting blocks grow as large as the fastest validators can handle while slower ones use its “skip vote” feature to abstain. Critics warned the change could favor well-funded operators and increase centralization.
