Solana just changed how its validators agree a transaction can’t be undone, and the change is structural rather than cosmetic. The network replaced TowerBFT, its original consensus protocol, with a new system called Alpenglow that moves validator votes off the chain entirely, collapsing a two-step confirmation process into one.

The swap happened live, on the running network, without a restart. Validators carried the existing ledger forward through the handoff, according to Decrypt. Anza, Solana’s core engineering team, says the result cuts the time it takes a transaction to become irreversible from about 12.8 seconds to a target of roughly 150 milliseconds.

Key Takeaways

  • Alpenglow activated at slot 459,653,000 around 14:00 UTC on Wednesday, retiring TowerBFT, which had run Solana since its March 2020 launch, according to The Block.
  • The protocol targets 150-millisecond finality, down from about 12.8 seconds, by moving validator votes off-chain and bundling them into certificates.
  • 654 validators took part in a switch that took less than a second, Anza said, though the Solana Foundation’s own dashboard showed 6.6-second finality shortly after, per Decrypt.
  • Further changes, including Rotor for block propagation and a post-quantum design called Quantumglow, are planned with no committed dates.
  • SOL rose about 2% to near $108 on the news, in line with Bitcoin, while US Solana ETFs stayed in outflows for a 13th day this month, according to CoinMarketCap.

How the Vote Mechanism Changes

TowerBFT required validators to write their votes into later blocks on the chain itself, which is why full finality took roughly 12.8 seconds even though Solana gave a conditional confirmation within seconds. Alpenglow removes that step. Validators now send votes directly to each other and bundle them into certificates, weighted by each validator’s stake, according to The Block.

Anza’s head of research, ETH Zurich professor Roger Wattenhofer, described the fast path in an interview cited by Decrypt: “If you have 80% of the votes say yes, then it’s immediately done. You immediately know that the block is final.” A second route needs two rounds of voting at 60% of stake each, which Wattenhofer said can be quicker for validators clustered close together, such as in Europe.

The protocol is built to keep functioning with up to 20% of stake acting maliciously and another 20% offline. If the network goes 10 seconds without finalizing a block, validators gradually lengthen block times until messages get through, Wattenhofer said: “Eventually, even a bad internet will give us a block again.”

The Activation Itself

The handoff took place at slot 459,653,000, with Anza CEO Brennan Watt calling it “the largest, most ambitious blockchain upgrade since ‘The Merge,’” adding that Anza took the design “from a white paper to running on mainnet in under 18 months,” per The Block. Anza’s research team, led by Wattenhofer along with Quentin Kniep and Kobi Sliwinski, published the Alpenglow design in May 2025, and validators had voted 98% in favor of adopting it in September 2025, according to Decrypt.

Anza said the design went through red teaming, penetration testing, repeated migration rehearsals and a security competition before moving through testnet and devnet, then onto mainnet with 654 validators taking part in a switch Anza said took less than a second on X, per CoinMarketCap. That figure is still an early read: the Solana Foundation’s own dashboard showed 6.6-second finality shortly after the switch, as Cointelegraph reported and Decrypt and CoinMarketCap both relayed, meaning the 150-millisecond figure remains a design target rather than a confirmed live measurement.

Why Traders Aren’t Fully Sold

Faster finality matters most to exchanges crediting deposits and merchants waiting for a payment to clear, Anza said, but it doesn’t change the competitive dynamics inside trading. Robin Nordnes, who runs Solana trading-infrastructure firm Raiku, told CoinDesk, as cited by CoinMarketCap: “For desks, faster finality is nice, but it was never the bottleneck. Being close to where the block is built still is.”

That skepticism lines up with the token’s recent flows. US spot Solana ETFs posted net outflows in 13 of 14 sessions this month through Tuesday, with Bitwise’s fund accounting for roughly three-quarters of the money pulled and total fund assets down about 16% from their late-September peak, per CoinMarketCap. SkyAI, formerly Sharps Technology, also sold its entire SOL position to fund a share buyback, as The Block reported Monday. SOL still rose about 1.8% to near $108.40 on Wednesday, a move CoinMarketCap said roughly matched Bitcoin’s 2.2% gain over the same period, rather than reflecting the upgrade alone.

Alpenglow doesn’t touch how staking rewards accrue, Galaxy noted, per CoinMarketCap. Validators now pay a flat fee of about 0.8 SOL a day each, roughly 500 SOL a day across the network, which the research firm described as a small cost relative to overall validator economics.

What to Watch

Anza has outlined, but not scheduled, several follow-on changes. Rotor would alter how block data spreads across the network, and geographically aware leader rotation is meant to cut delays as block production passes between validators, according to The Block. Quantumglow, a research design, would swap Alpenglow’s elliptic-curve cryptography for hash-based signatures to resist quantum computers while keeping the same finality speed. Separately, Jump Crypto’s Firedancer team proposed SIMD-0370 in September 2025, which would let Solana’s block capacity scale with validator hardware instead of staying capped at the current 60 million compute-unit limit. None of these carries a committed release date, and whether live finality readings converge toward the 150-millisecond target, rather than the 6.6-second figure seen shortly after activation, is the first thing to track.

Sources