Solana Alpenglow: the biggest consensus overhaul in Solana's history

Last updated: April 7, 20266 min read

Solana Alpenglow consensus upgrade explainer

Solana transactions take 12.8 seconds to finalize. After Alpenglow, that drops to 150 milliseconds.

But speed is just the surface. Alpenglow tears out Proof of History and Tower BFT, the two systems that have defined Solana's consensus since launch, and replaces them with a simpler architecture that also eliminates on-chain vote transactions. Those votes currently consume ~75% of Solana's block space. It passed governance with 98.27% approval, it's in private cluster testing now, and mainnet is expected late 2026.

What is Alpenglow?

Alpenglow is a complete replacement of Solana's consensus layer. The new architecture finalizes transactions in 100-150 milliseconds instead of 12.8 seconds, and by moving validator voting off-chain, it frees roughly three-quarters of block space for actual user transactions.

Formalized as SIMD-0326, the upgrade is built around two new protocols:

Votor: a new voting and finalization system that replaces Tower BFT
Rotor: a new block propagation system that replaces Turbine

Think of it this way: Tower BFT is like a 32-step approval process where each block climbs through layers of confirmation. Alpenglow collapses that into one or two quick rounds. The upgrade also retires Proof of History, Solana's original cryptographic clock. Instead of running a continuous hash chain, validators use a fixed 400ms block time with local timeouts.

Who's building it?

The research came from Professor Wattenhofer's distributed systems lab at ETH Zurich. Anza, the core team behind the Agave validator client, is leading implementation. Jump Crypto's Firedancer team is collaborating on multi-client compatibility.

Where does it stand right now?

May 2025: Unveiled at Solana Accelerate in New York
September 2025: Governance vote passed, 98.27% yes, 52% of total stake participated
Early 2026: Available on Agave's master branch for private cluster testing
Q3 2026: Full release targeting Agave 4.1
Late 2026: Mainnet activation expected after community testing and security audits (not yet on production clusters as of April 2026)

How does Alpenglow work?

Three changes.

1. Votor: how blocks get finalized

The current system (Tower BFT) requires 32 incremental confirmations. Each layer adds a lockout period.

Votor replaces this with one or two rounds of voting:

Fast path (~100ms): If 80%+ of validators approve in the first round, the block is final. Done.
Slow path (~150ms): If 60-80% approve in round one, a second round runs. If 60%+ approve again, it's final.

Both paths run simultaneously. Whichever finishes first wins.

Validators send votes as lightweight UDP messages directly to each other, not as on-chain transactions. Votor bundles these votes using BLS signature aggregation: thousands of individual signatures compressed into one compact proof. Only the aggregated certificate (~1,000 bytes) lands on-chain, replacing the ~500KB of vote data currently recorded per slot.

This is a big deal. Right now, ~75% of all Solana transactions are validator votes. Alpenglow eliminates them entirely.

2. Rotor: how block data spreads

Note: Rotor is part of the broader Alpenglow vision but ships as a separate SIMD from the core Votor consensus changes.

Currently, Solana uses Turbine, a multi-layer relay tree with a 200-node fanout. Block data hops through multiple layers. Think of it like a game of telephone.

Rotor replaces this with a single-hop broadcast: the block producer sends data to a small set of relay nodes, and those relays push it to everyone at once. From a phone tree to a group broadcast.

The numbers from the Alpenglow whitepaper: transmitting 1,500 shreds takes 18ms on 1 Gb/s bandwidth. Reaching 80% of total stake needs only ~150 nodes in about 2ms.

3. No more Proof of History

Alpenglow retires Solana's cryptographic clock. Instead, validators use a fixed 400ms block time and local clocks with timeouts:

Block arrives before timeout: approve it (NotarVote)
Timeout expires with no block: skip it (SkipVote)

The protocol tolerates clock drift proportionally: 5% drift only needs a 5% timeout extension.

How does the security model change?

Alpenglow introduces a "20+20" fault tolerance framework:

Scenario	Traditional BFT	Alpenglow
Malicious validators	Up to 33%	Up to 20%
Offline validators	Up to 33%	Up to 20%
Both at once	Up to 33%	Up to 40%

Why does this work? For safety: conflicting forks can't both reach quorum because 80% + 60% = 140% > 100%. For liveness: the network still finalizes through the slow path even if 20% of stake goes offline.

The trade-off: pure Byzantine tolerance drops from 33% to 20%. But in practice, real-world failures are mixed. Some nodes are malicious, some are just offline. The combined 40% tolerance handles that scenario better than traditional BFT's 33%.

What are the performance gains?

Metric	Current Solana	With Alpenglow
Transaction finality	~12.8 seconds	100–150ms
Optimistic confirmation	500–600ms	100–150ms (deterministic)
Max theoretical TPS	~65,000	~107,000
Vote transactions (% of block space)	~75%	0%
Validator vote cost	~1 SOL/day	Eliminated
Min. profitable stake	~4,850 SOL	~450 SOL
On-chain vote data per slot	~500KB	~1,000 bytes

Simulation results show 65% of stake finalizes within 50ms of raw network latency. With a ~70ms longest one-way network hop, fast-path finality lands around 120-150ms.

What changes for validators?

The economics shift:

Vote fees disappear. Currently ~1 SOL/day per validator. Alpenglow moves all voting off-chain.
New cost: Validator Admission Ticket (VAT). ~0.8 SOL/day, entirely burned to reduce inflation.
Validator cap: 2,000 (highest stake selected).
Barrier to entry drops ~90%. Minimum profitable stake goes from ~4,850 SOL to ~450 SOL.
Operating costs drop 20-50% overall.
HSM support. Identity keys can live in hardware security modules since they no longer sign high-frequency vote transactions.

What changes for developers and users?

For developers, the biggest change isn't the speed number. It's that finality becomes a single thing.

Today, Solana has three commitment levels: processed, confirmed (~500ms, probabilistic), and finalized (~12.8s, deterministic). Every developer makes a choice per feature: accept rollback risk and show the user a fast result, or wait 12.8 seconds and be certain. Exchange deposit crediting, bridge confirmations, and liquidation engines all wait for finalized because even a small rollback risk is unacceptable. Everything else uses confirmed and hopes for the best.

Alpenglow collapses this into one level. 100-150ms, deterministic finality on every transaction. No commitment level trade-offs. No retry logic for disappeared transactions. One check, done.

That's what makes new patterns viable: onchain order books can offer tighter spreads because market makers know their quote state is final in 150ms. Perpetual exchanges might see lower latency costs on JIT auctions. Cross-chain bridges eliminate the Solana-side finality bottleneck. And exchange deposit crediting goes from a 13-second wait to instant.

For users, transactions confirm fast enough to feel instant. The bigger shift is capacity: removing vote transactions frees up 75% of block space, which means lower fees and less congestion during high-traffic periods.

For financial institutions building on Solana, deterministic sub-second finality closes the gap with traditional payment rails. Stablecoin settlement, tokenized treasury operations, and cross-border transfers no longer need a 13-second buffer.

What's still coming?

Alpenglow also makes life easier for Firedancer, Jump Crypto's independent validator client written in C/C++. Firedancer improves single-node performance; Alpenglow redesigns how nodes reach consensus. The simpler protocol means less consensus complexity to reimplement in a second codebase — a practical win for multi-client diversity.

SIMD-0326 covers only the Votor consensus changes. Three components from the Alpenglow whitepaper are deferred to future proposals:

Rotor (block propagation), separate SIMD
Smart Sampling (bandwidth optimization)
Lazy/Asynchronous Execution

The path to mainnet: Agave 4.1 release in Q3 2026, community testing and security audits through Q4, mainnet activation late 2026.

Build on Solana with Alchemy

Alpenglow makes Solana faster, but you still need reliable infrastructure to build on it.

Alchemy's Solana platform gives you archival data access up to 20x faster than other providers, gRPC streaming at half the price with 6,000-block reconnection recovery, and 10x faster heavy method calls like getProgramAccounts.

Smart WebSockets that eliminate dropped connections, plus staked connections for reliable transaction routing
Gasless transaction support to remove friction for end users
99.99% reliability with multi-region deployment and 3-5 layers of autonomous failover

Solflare, Robinhood, OpenSea, and Circle already run on Alchemy's Solana infrastructure. Pay-as-you-go pricing, no long-term commitments.

Get started with a free Solana RPC endpoint →

Frequently asked questions

What is Solana Alpenglow?

Alpenglow is a complete replacement of Solana's consensus layer that reduces transaction finality from 12.8 seconds to 100-150 milliseconds and eliminates on-chain vote transactions, freeing roughly 75% of block space for user transactions.

What does Alpenglow replace in Solana's current architecture?

Alpenglow removes Proof of History, Tower BFT, and on-chain vote transactions, replacing them with Votor (a new voting system), Rotor (a new block propagation system), and a fixed 400ms block time with local timeouts.

When will Alpenglow launch on mainnet?

Alpenglow is expected to activate on mainnet in late 2026, following the Agave 4.1 release in Q3 2026 and subsequent community testing and security audits through Q4 2026.

How does Alpenglow achieve faster finality?

Votor collapses the current 32-step confirmation process into one or two rounds of voting: the fast path finalizes at ~100ms with 80%+ validator approval, while the slow path finalizes at ~150ms with 60%+ approval across two rounds.

How does Alpenglow impact validator costs?

Alpenglow eliminates the ~1 SOL/day cost for on-chain vote transactions and reduces the minimum profitable stake from ~4,850 SOL to ~450 SOL, though validators will pay a new ~0.8 SOL/day Validator Admission Ticket fee that is burned.

What are the security trade-offs in Alpenglow?

Alpenglow tolerates up to 20% malicious validators and 20% offline nodes (or 40% combined), compared to traditional BFT's 33% limit, using an 80% quorum for fast-path finality and 60% for slow-path finality to prevent conflicting forks.

How does Alpenglow change commitment levels for developers?

Alpenglow collapses Solana's three commitment levels (processed, confirmed, finalized) into a single deterministic finality at 100-150ms, eliminating the need to choose between speed and rollback risk.

What is the current status of Alpenglow development?

Alpenglow passed governance with 98.27% approval in September 2025, is available on Agave's master branch for private cluster testing as of early 2026, and is not yet on production clusters as of April 2026.

Key resources

Alchemy Newsletter

Be the first to know about releases

Sign up for our newsletter

Get the latest product updates and resources from Alchemy

Over 80,000 subscribers

By entering your email address, you agree to receive our marketing communications and product updates. You acknowledge that Alchemy processes the information we receive in accordance with our Privacy Notice. You can unsubscribe anytime.