Short answer: Proof of History (PoH) is the cryptographic clock Anatoly Yakovenko designed in 2017 for what became Solana: a chain of SHA-256 hashes, each computed from the one before, that lets a block producer stamp transactions with a provable order and a measured passage of time. It was never a consensus mechanism on its own; it was the clock for Tower BFT, Solana’s proof-of-stake voting. On October 21, 2026, at about 13:54 UTC, Alpenglow’s Votor replaced Tower BFT at slot 459,653,000. Proof of History no longer paces block production, and Solana’s finality, about 12.8 seconds under Tower BFT, now has a target of about 150 milliseconds.
Key takeaways
- Proof of History orders events; it does not decide which blocks are final. Solana’s whitepaper presents it as something used “alongside a consensus algorithm” to cut the messages validators must exchange (Solana whitepaper, v0.8.13).
- Solana stopped using Proof of History as its clock on October 21, 2026, when Alpenglow’s first phase, Votor, took over consensus at slot 459,653,000, at about 13:54 UTC (Solana’s public RPC); leaders no longer emit PoH ticks (Solana Foundation).
- Votor finalizes a block in one round when 80 percent of stake votes, or in two rounds with 60 percent. Anza’s simulations put median finality at about 150 milliseconds, against about 12.8 seconds under Tower BFT (Anza; Solana Foundation). The Foundation’s mainnet dashboard put 95th-percentile observed finality at about 0.34 to 0.51 seconds through October 26, 2026.
- Votes stopped being transactions, so headline TPS fell. Solana’s public RPC counted about 5,360 transactions per second in the half hour before the switch, 57 percent of them votes, and about 2,000 per second in the 12 minutes after, with the vote count at or near zero.
- As of October 2026, validators pay a Validator Admission Ticket of 0.8 SOL per epoch, burned, instead of about 1 SOL per epoch in vote fees, and Rotor, Alpenglow’s second half, is not yet scheduled.
| Question | Tower BFT era (2020 to October 21, 2026) | Alpenglow’s Votor (from slot 459,653,000) |
|---|---|---|
| What paces block production? | Proof of History: the leader’s hash chain and its tick entries | Votor’s local timeouts and vote certificates; leaders stop emitting ticks |
| How do votes travel? | As vote transactions inside blocks | As messages sent directly between validators, aggregated into certificates |
| When is a block final? | About 12.8 seconds | Target about 150 milliseconds (median, simulated); sometimes as fast as about 100 milliseconds |
| How much stake must vote? | Stake-weighted votes with lockouts that double with each new vote | 80 percent in one round, or 60 percent in two rounds, run at the same time |
| What does voting cost a validator? | About 1 SOL per epoch in vote fees | A flat 0.8 SOL per epoch, burned (the Validator Admission Ticket) |
| How is block data broadcast? | Turbine | Still Turbine; Rotor is not yet scheduled |
| How long is a slot? | 400 milliseconds for years, cut to 200 in four stages that finished in October 2026 | About 200 milliseconds; slots still start on a fixed cadence |
| What do developers wait for? | “confirmed” and “finalized” are different levels | “confirmed” and “finalized” mean the same; “confirmed” will be removed later |
In this article: What is Proof of History? | Who invented it? | Is it a consensus mechanism? | Does Solana still use it? | What changed on October 21, 2026? | Finality now | Why TPS dropped | Votor and Rotor | Validators | FAQ
What is Proof of History?
Proof of History is a verifiable record of time built from a hash chain. A computer runs SHA-256 again and again, feeding each output back in as the next input and counting the steps. Because every step needs the result of the step before, nobody can produce hash number one million without computing the 999,999 that come first.
That serial chain becomes a timestamp service. When a transaction, or a hash of it, is mixed into the chain at some position, the record shows that the transaction existed before every hash that follows. Solana’s whitepaper, “Solana: A new architecture for a high performance blockchain,” calls PoH “a proof for verifying order and passage of time between events.” Solana’s glossary puts it more plainly: a stack of proofs, each showing that some data existed before the proof was created.
Producing the chain cannot be sped up by adding processors, but checking it can. The whitepaper’s example splits a long sequence into 4,000 slices so that a GPU with 4,000 cores can check every slice at once, each from its starting hash to its last. Generation is one long thread; verification is thousands of short ones.
How Proof of History worked inside Solana
Solana gives each slot to one validator, the leader, according to a published leader schedule. Until October 21, 2026, the leader ran the PoH chain during its slots, wove incoming transactions into it and emitted “ticks,” which Solana’s glossary defines as ledger entries that estimate wall-clock duration. Other validators could replay the chain, see how much time had passed and know when a leader’s turn was over without asking anyone.
Tower BFT, Solana’s voting protocol, used the same clock. Its lockouts, the periods during which a validator that has voted for a fork may not vote for a rival one, are measured in slots, and each new vote doubles the lockouts of the votes before it. Anza’s documentation says a vote reaches maximum lockout at 32 votes, and Solana’s glossary defines finality as the point when validators holding two-thirds of stake share a common root.
Why did Solana use Proof of History?
To cut the messages validators had to exchange. With a shared clock in the ledger, validators could tell when a leader’s turn was over and when each vote was cast without asking one another, and Solana’s whitepaper says this can reduce the messaging overhead of reaching agreement.
Is Proof of History a verifiable delay function?
Only loosely. Anza’s Tower BFT documentation calls Proof of History a verifiable delay function, and Solana’s glossary says that, like one, it can be verified in less time than it took to produce. That works as shorthand. The formal definition, published by Dan Boneh, Joseph Bonneau, Benedikt Bünz and Ben Fisch at CRYPTO 2018, asks for a function that needs a set number of sequential steps to evaluate but can be checked quickly. A PoH chain meets the first test. Checking it takes about as much total computation as producing it, and it is fast only because the work can be spread across many cores.
Who invented Proof of History?
Anatoly Yakovenko, Solana’s co-founder, invented Proof of History in 2017. Fortune’s 2022 profile says he had spent nearly 13 years at Qualcomm working on distributed systems, and that the idea came on a night when, after two coffees and a beer, he stayed awake until 4 a.m.
The wireless background matters. Many cellular systems share a radio channel through Time-Division Multiple Access (TDMA), which gives each device its own time slot on a shared clock. In ARK Invest’s FYI podcast of October 30, 2025, the episode notes describe one part of the conversation as time as a cryptographic primitive, solving blockchain inefficiency with TDMA. Solana’s leader schedule, which hands out slots to validators in turn, follows the same pattern.
He did not build it alone. Fortune names Greg Fitzgerald and Stephen Akridge, fellow Qualcomm alumni, among his co-founders, and says Raj Gokal joined as a co-founder at the end of 2017. The network produced its first block on March 16, 2020. The fuller story of those years is in the complete history of Solana on this site.
The Prologue of LIGHT tells the night of the idea this way:
Blockchains, Toly realized, had no clock. […]
The proposed solutions (sharding the blockchain into parallel chains, moving transactions off-chain to Layer 2 networks, building elaborate bridges between separate systems) struck Toly as baroque. They were patches on top of patches, adding complexity without addressing the root cause. The root cause was that every node in a blockchain network spent most of its time doing the same thing: arguing with every other node about the order in which events had occurred.
What if you didn’t need to argue? What if time was already in the data?
[…] Take a piece of data. Hash it with SHA-256, the same cryptographic function that secures Bitcoin. Take the output and hash it again. And again. And again. Each hash depends on the one before it, which means the chain can only be computed sequentially. You cannot skip ahead. You cannot parallelize it. And because the computation takes a known amount of real time, anyone can verify, after the fact, that a specific duration must have elapsed between any two hashes in the chain.
This chain of hashes was a clock. Not a physical clock, which can be spoofed or manipulated, but a cryptographic clock, a mathematical proof of the passage of time, embedded in the data itself. If you stamped every transaction with its position in this hash chain, you didn’t need validators to negotiate ordering. The order was proven. The consensus was in the math.
Toly called it Proof of History.
Excerpted from LIGHT: The Solana Story and the Man Who Turned Time Into Code by Ryan Bethencourt, Prologue, licensed CC BY-SA 4.0.
In the running network, the hash chain settled the order of transactions, and validators still voted on which blocks to keep. That voting layer, and the clock it leaned on, is what Alpenglow replaced.
Is Proof of History a consensus mechanism?
No. Consensus decides which version of the ledger the network accepts as final; Proof of History only puts events in a provable order and spacing. Solana’s whitepaper says PoH works “alongside a consensus algorithm such as Proof of Work (PoW) or Proof of Stake (PoS)” and that it reduces the messaging overhead of reaching agreement.
On Solana, that consensus was Tower BFT, a proof-of-stake protocol in the Byzantine fault tolerant family. Validators voted with weight in proportion to their stake, and PoH gave every vote a shared sense of when it was cast. The clock made voting cheaper. It never replaced the vote.
People often list PoH beside Proof of Work and Proof of Stake as if the three were rivals. They answer different questions:
| Mechanism | The question it answers | Where it runs |
|---|---|---|
| Proof of Work | Who may add the next block, by spending computation | Bitcoin |
| Proof of Stake | Whose votes count, weighted by tokens at stake | Ethereum since September 15, 2022; Solana |
| Proof of History | In what order did events happen, and how much time passed between them | Solana, as its block-pacing clock until October 21, 2026 |
| Tower BFT | Which fork is final, with PoH as the clock for lockouts | Solana until October 21, 2026 |
| Votor (Alpenglow) | Which block is final, using local timeouts and stake-weighted certificates | Solana from October 21, 2026 |
For a side-by-side of how Solana and Ethereum differ beyond consensus, see Solana vs Ethereum in 2026.
Does Solana still use Proof of History?
No. Since Votor took over at slot 459,653,000 on October 21, 2026, Solana no longer uses Proof of History to pace blocks or to time consensus. The table of breaking changes on the Solana Foundation’s Alpenglow page says: “PoH is gone, and tick entries go with it.” The same page states that “Proof of History no longer paces block production, so tick entries leave the shred stream,” and that Votor’s timeouts and certificates pace block production instead.
Three things change in the ledger, by the Foundation’s account:
- Ticks no longer keep time. Leaders stop emitting the 64 evenly spaced tick entries that used to mark each slot, so any tool that counted ticks to measure time inside a slot has to stop.
- Time is kept per slot. The ledger keeps its slot-level sense of time and loses the finer, within-slot one. Slots still start on a fixed cadence.
- Block timestamps come from the leader. Each leader sets its block’s timestamp within a bound: no earlier than the parent block’s timestamp and no later than the parent’s plus twice the elapsed slot duration. Over an honest majority, the Foundation says, the clock converges on real time.
The authors of SIMD-0326, the proposal behind Votor, describe its timeouts as serving a purpose akin to Proof of History, but without any notion of synchronized time. Each validator runs its own timers, so clock drift between machines has no protocol meaning. When Anza unveiled Alpenglow on May 19, 2025, it named TowerBFT and Proof of History as the legacy components it was saying goodbye to.
Yakovenko has raised the possibility himself. ARK’s notes for the October 2025 episode list a segment, at 11:33, on why Proof of History may disappear and what replaces it. LIGHT’s Chapter Fifteen reads the remark like this:
What he meant was not that the innovation was wrong or that it had failed. What he meant was that Proof of History was a means to an end, not the end itself. The purpose of PoH was to eliminate the communication overhead of consensus by embedding a trustworthy timeline in the data. If future consensus mechanisms (Alpenglow, or something not yet invented) could achieve the same result through different techniques, then PoH might fade into the background, absorbed into the infrastructure it enabled, no longer a visible or distinct feature of the system.
Excerpted from LIGHT: The Solana Story and the Man Who Turned Time Into Code by Ryan Bethencourt, Chapter Fifteen, licensed CC BY-SA 4.0.
What survives is the design PoH made possible: one global ledger, leaders taking turns in short fixed slots, and transactions processed in parallel. The clock that once timed every slot has handed that job to Votor’s timeouts and certificates.
What changed on October 21, 2026?
Alpenglow’s first phase went live on Solana mainnet: Votor replaced Tower BFT at slot 459,653,000, at about 13:54 UTC.
Its feature gate (A1pengvuM6JEcyNuTnMqepBKhwHE3N6PmUrdATGawhJS) activated at slot 459,648,000, the first slot of epoch 1064: the feature account on-chain records that slot, and the Solana Foundation’s Alpenglow page, marked “Live on Mainnet” when read on October 26, 2026, confirms activation at the start of epoch 1064, at approximately 13:36 UTC. Tower BFT kept running for 5,000 more slots, about 18 minutes by the Foundation’s account, up to the handover. Solana’s public RPC records the last Tower BFT block at slot 459,652,999, stamped 13:54:00 UTC, and the first block after the handover at slot 459,653,008, stamped 13:54:03 UTC.
The eight slots in between have no block: under the migration rules in SIMD-0384, validators certify the last Tower BFT block before the boundary as Alpenglow’s genesis block and discard anything built after it. Mainnet’s genesis certificate, which the getAgGenesisCert RPC method returns once a cluster has migrated, names slot 459,652,999; the Foundation counts votes from 446 of the network’s 654 validators in it. From slot 459,653,008 on, validators’ consensus votes no longer appear in blocks as transactions.
| Date | Event | Source |
|---|---|---|
| May 19, 2025 | Anza researchers Quentin Kniep, Kobi Sliwinski and Roger Wattenhofer unveil Alpenglow, calling it “the biggest change to Solana’s core protocol since, well, ever” | Anza |
| July 25, 2025 | SIMD-0326, the Votor proposal, is created | SIMD-0326 |
| September 2, 2025 | The validator vote on SIMD-0326 passes: 98.27 percent yes, 1.05 percent no, 0.69 percent abstain, with 52 percent of stake voting | Solana Status on X, as quoted by Bitcoinist |
| December 12, 2025 | Jump Crypto’s Firedancer client goes live on mainnet | The Block |
| July 8, 2026 | BLS public key management in vote accounts (SIMD-0387) activates on mainnet | Solana Foundation |
| July 22, 2026 | The Validator Admission Ticket (SIMD-0357) activates on mainnet | Solana Foundation |
| September 25, 2026 | Alpenglow activates on devnet (epoch 1167, 19:30 UTC) | Solana Foundation |
| October 2026 | Slot time reaches 200 milliseconds, down from 400, after four stages | Solana Foundation |
| October 13, 2026 | Anza announces October 21 as mainnet day | Solana Compass |
| October 21, 2026 | The feature gate activates at slot 459,648,000, the start of epoch 1064 (13:36 UTC); Tower BFT runs 5,000 more slots; Votor takes over at slot 459,653,000, and the first block after the handover, slot 459,653,008, is stamped 13:54:03 UTC | Solana Foundation; feature account and block times from Solana’s public RPC |
| Not yet scheduled | Rotor, Alpenglow’s data-broadcast half | Solana Foundation, as of October 2026 |
Advance estimates of the clock time, about 13:20 UTC on the Foundation’s page before the switch and about 14:50 UTC from Anza’s October 13 announcement on X as Solana Compass reported it, both missed; the slot number was the fixed point.
The chain kept producing blocks through the switch. By the same RPC’s block list, 292 of the first 300 slots after the handover produced blocks, against 96 of the 100 slots before it. Whether Solana has halted since is tracked in the dated guide to every Solana outage (publishing December 17).
For users, nothing about signing or sending a transaction changed. The Foundation’s guidance was that wallets and applications keep working as before, and that an application which only sends transactions and reads account state has nothing to migrate. For a sibling explainer on another chain’s 2026 protocol upgrade, see Zcash NU7 explained.
How long does Solana finality take now?
Votor targets a median finality of about 150 milliseconds, and Anza says finality can sometimes come as fast as 100 milliseconds. Under Tower BFT, finality took about 12.8 seconds, by the Foundation’s figure. The 400 milliseconds often quoted for Solana was its slot time, not its finality.
Those targets come with a caveat Anza states itself: the 150 and 100 millisecond figures come from simulations that used mainnet’s stake distribution in 2025 and leave out computation overhead. The Foundation’s Alpenglow transition dashboard does measure mainnet: it charts 95th-percentile finality as observed over RPC, with polling and network delay included. Read on October 26, 2026, every half-hourly point on its chart after the switch fell between about 0.34 and 0.51 seconds. That is a different measure from Anza’s simulated median.
Three numbers are easy to mix up:
| Number | What it measures | Value, as of October 2026 |
|---|---|---|
| Slot time | How often a new leader gets a turn to produce a block | About 200 milliseconds |
| Finality under Votor | How long until a block cannot be reverted | About 150 milliseconds median, sometimes as fast as 100 milliseconds (Anza’s simulations) |
| Finality under Tower BFT | The same, before the switch | About 12.8 seconds |
For developers, the change shows up in commitment levels. Under Tower BFT, “confirmed” (a supermajority of stake had voted for the block) arrived well before “finalized.” Under Votor the two mean the same thing, and the Foundation says “confirmed” will be deprecated and removed in a later release. Its advice was to point new code at “finalized” once Alpenglow went live, and not before, because on Tower BFT “finalized” still meant the 12.8-second wait.
Why did Solana’s TPS drop after Alpenglow?
Because validators’ votes stopped being transactions. Under Tower BFT, every vote was a transaction recorded in a block, and explorers counted it in transactions per second (TPS). Under Votor, votes travel as messages between validators, not as transactions, so the headline count falls while user activity does not. The Foundation told dashboard operators to “re-baseline transaction counts and TPS dashboards.”
Solana’s own counters show the change. Its public RPC reports transactions in one-minute samples, split into vote and non-vote transactions. Read on October 21, 2026, they give this picture:
| Window, October 21, 2026 (UTC) | Consensus | Transactions counted per second | Votes per second | Non-vote per second | Votes as a share |
|---|---|---|---|---|---|
| 13:23 to 13:53 (30 samples) | Tower BFT | About 5,360 | About 3,030 | About 2,330 | 57 percent (48 to 64 percent by minute) |
| 13:54 to 14:06 (12 samples) | Votor | About 2,000 | 0 | About 2,000 | 0 percent |
The headline count fell by about 63 percent within minutes, and most of the fall was the vote column dropping to zero or close to it, while non-vote traffic stayed in a similar range. Block by block the change is just as plain: the last Tower BFT block, slot 459,652,999, held 1,051 transactions, 649 of them votes, and the first block after the handover, slot 459,653,008, held no votes among its 1,701 transactions.
The number to compare across the switch is non-vote TPS. Alchemy’s explainer, updated October 21, 2026, put votes at about 75 percent of Solana’s transactions before the switch; the share moves with load, because the vote count stays fairly steady while user traffic rises and falls. A falling total after October 21 says nothing about demand on its own.
What are Votor and Rotor?
Votor and Rotor are the two halves of Alpenglow. Votor is the voting and finality protocol, which took over from Tower BFT on October 21, 2026. Rotor is a new way to broadcast block data that is meant to replace Turbine, and as of October 2026 it is not yet scheduled.
Votor. Validators send votes directly to each other rather than through gossip or on-chain transactions. Once enough stake has voted, the votes can be combined into a certificate, an aggregate BLS signature that stands for all of them. Votor runs two paths at the same time: a block is final after one round if 80 percent of stake votes for it, or after two rounds if 60 percent does, whichever finishes first. Anza says the design tolerates up to 20 percent adversarial stake plus another 20 percent that is offline or unresponsive.
Rotor. Turbine, Solana’s current broadcast system, passes pieces of each block through a multi-layer tree of validators. Rotor uses a single layer of relay nodes, which cuts the number of network hops, and draws on each node’s bandwidth in proportion to its stake. SIMD-0326 left Rotor out of scope: it gets its own proposal, and Turbine stays in place until then.
The designers are Quentin Kniep, Kobi Sliwinski and Roger Wattenhofer at Anza, the Solana Labs spin-off that builds the Agave validator client. Cointelegraph’s report on the proposal noted that switching to Alpenglow would not completely shield Solana from outages.
What changes for validators?
Validators stopped paying vote transaction fees and started paying a flat Validator Admission Ticket (VAT) of 0.8 SOL per epoch, burned from the vote account. For the migration they also had to run an Agave-based client and register a BLS public key on-chain. The groundwork went in months earlier: BLS key management (SIMD-0387) activated on mainnet on July 8, 2026 and the Validator Admission Ticket (SIMD-0357) on July 22, 2026, and validators without a registered BLS key were excluded from the admitted set and stopped taking part in consensus, the Foundation says.
| Item | Under Tower BFT | Under Alpenglow, as of October 2026 |
|---|---|---|
| Cost of voting | About 1 SOL per epoch in vote transaction fees, paid from the identity account | 0.8 SOL per epoch, taken from the vote account and burned; the first fee was charged at the start of epoch 1064 and paid for epoch 1065 |
| If the fee cannot be paid | Not applicable | The validator is not admitted for the next epoch: it does not vote or produce blocks, and loses that epoch’s rewards |
| Client during the migration | Agave, Jito-Solana, Firedancer, Frankendancer and others | An Agave-based client (Agave, Jito-Solana or Harmonic Salsa, v4.3.2) before epoch 1064; Firedancer and Frankendancer were not supported during the migration; Firedancer’s v26.10.2 (October 14, 2026) supports Alpenglow once the switch is done |
| Keys | Votes signed as ordinary transactions | A BLS public key registered on-chain (SIMD-0387); validators without one are excluded from admission |
| Commission | By default, inflation rewards go to the vote account and block revenue to the identity account | Same defaults; a separate change, SIMD-0232, lets a validator route block revenue to the vote account to help fund the fee, two epochs after the change is made |
The client rule touched a sizable share of stake. On October 13, 2026, Solana Compass found that validators among the 500 largest whose clients were labeled as Firedancer variants held about 15.8 percent of all activated stake, and the Foundation told Firedancer operators to switch to an Agave-based client before epoch 1064. Full Firedancer had gone live on mainnet on December 12, 2025. Its v26.10.2 release of October 14, 2026, a mainnet-ready build, added Alpenglow support but said “Migration from Tower to Alpenglow will not be supported,” so Firedancer operators were told to run an Agave-based client through the handover and could return afterward. The site’s guide to Firedancer and Alpenglow covers that return and why Solana wanted a second client.
The fee also changes validator arithmetic. The Foundation warns that commission income is not guaranteed to cover the VAT, and that rewards arriving at an epoch boundary cannot rescue an account that is already short when admission is checked. Its advice: fund the first fee directly, then keep a buffer above the rent-exempt minimum plus the next fee.
Frequently asked questions
What does PoH mean in crypto?
PoH stands for Proof of History, the hash-chain clock Anatoly Yakovenko designed in 2017 for what became Solana. A leader runs SHA-256 in a continuous sequence and mixes transactions into it, so the record proves their order and roughly how much time passed between them. Since October 21, 2026, Solana no longer uses it to pace blocks.
Is Proof of History the same as Proof of Stake?
No. Proof of Stake decides whose votes count, weighted by tokens at stake. Proof of History only orders events and measures time between them. Solana used both: Tower BFT, its proof-of-stake consensus, used PoH as its clock until Alpenglow’s Votor replaced Tower BFT on October 21, 2026.
Did Alpenglow remove Proof of History from Solana?
Yes, as Solana’s clock: it removed Proof of History from consensus and from pacing blocks. The Solana Foundation says PoH no longer paces block production, leaders stop emitting tick entries, and Votor’s timeouts and certificates set the pace instead. The ledger keeps time per slot, and each leader sets its block’s timestamp within a fixed bound.
What is a tick on Solana?
A tick is a ledger entry that estimates wall-clock duration. Under Proof of History, leaders emitted ticks inside each slot, so validators could tell how much time had passed. Under Alpenglow, leaders stop emitting ticks, and the Foundation tells developers to stop counting them.
Do Solana users need to do anything after Alpenglow?
No. The Solana Foundation’s guidance for users is “No action needed”: signing, sending and approving transactions do not change, and wallets and applications keep working. Because no action is needed, a site or message asking you to connect a wallet, swap tokens or sign a transaction “for Alpenglow” is not part of the upgrade. Developers who wait for finality should use the “finalized” commitment level, which under Votor means the same as “confirmed”.
When will Rotor launch?
No date has been set. As of October 2026, the Solana Foundation says Rotor is planned for a later release now that Votor is live, and is not yet scheduled. Until then, Solana keeps using Turbine to broadcast block data.
Who designed Alpenglow?
Quentin Kniep, Kobi Sliwinski and Roger Wattenhofer at Anza, the company behind the Agave validator client. Anza unveiled Alpenglow on May 19, 2025, and validators approved SIMD-0326, its Votor phase, in a vote that passed on September 2, 2025 with 98.27 percent of the votes cast in favor.
Sources and further reading
- Solana Foundation, “Alpenglow” upgrade page, updated October 2026, read October 21 and October 26, 2026; by October 26 it was marked “Live on Mainnet” and gave the activation time, the genesis block and its 446 certificate signers after the fact (also PoH and ticks, finality, the Validator Admission Ticket, client support, developer guidance, Rotor status). solana.com
- Solana Foundation, “Alpenglow Transition Dashboard,” mainnet chart “Observed finality latency (p95),” seven-day view, read October 26, 2026. solana.com
- Quentin Kniep, Kobi Sliwinski and Roger Wattenhofer, Anza blog post introducing Alpenglow, May 19, 2025 (Votor’s 80 and 60 percent paths, simulated 150 and 100 millisecond finality, Rotor, 20+20 resilience). anza.xyz
- SIMD-0326, “Alpenglow,” Solana Improvement Documents, created July 25, 2025 (scope, timeouts, certificates, Rotor out of scope). github.com
- SIMD-0384, “Alpenglow migration,” Solana Improvement Documents (the boundary 5,000 slots after activation, the genesis block and genesis certificate). github.com
- Anatoly Yakovenko, “Solana: A new architecture for a high performance blockchain,” whitepaper v0.8.13 (definition of PoH, use alongside a consensus algorithm, parallel verification). solana.com
- Solana documentation, “Terminology” (tick, slot, leader, Proof of History, confirmed block). solana.com
- Anza documentation, “Tower BFT” (time source, lockouts, maximum lockout at 32 votes). docs.anza.xyz
- Dan Boneh, Joseph Bonneau, Benedikt Bünz and Ben Fisch, “Verifiable Delay Functions,” CRYPTO 2018 (IACR ePrint 2018/601). eprint.iacr.org
- Anne Sraders and Declan Harty, Fortune profile of Anatoly Yakovenko, April 14, 2022 (Qualcomm years, the 2017 night, co-founders). fortune.com
- ARK Invest, FYI podcast, “Solana’s Vision For A Decentralized Future With Anatoly Yakovenko,” October 30, 2025 (episode notes: TDMA at 6:25; why Proof of History may disappear at 11:33). podcast.show
- Wikipedia, “Solana (blockchain platform)” (first block March 16, 2020). wikipedia.org
- Solana Status, post on X, September 2, 2025, as quoted by Bitcoinist in its report on the SIMD-0326 vote, last updated September 3, 2025 (98.27 percent yes, 1.05 percent no, 0.69 percent abstain; 52 percent of stake voting).
- Cointelegraph, report on Anza’s Alpenglow proposal, May 20, 2025 (Anza as a Solana Labs spin-off; the white paper on outages). cointelegraph.com
- The Block, report on Firedancer reaching Solana mainnet, December 12, 2025. theblock.co
- Firedancer, “Firedancer Mainnet Release v26.10.2,” GitHub, October 14, 2026 (Alpenglow support; migration from Tower BFT not supported). github.com
- Solana Compass, “Anza Sets Alpenglow Mainnet Date: Solana’s Consensus Switch Is Scheduled for October 21 at About 14:50 UTC,” October 13, 2026 (Anza’s announced time; Firedancer-family stake share). solanacompass.com
- Alchemy, “Solana Alpenglow: the biggest consensus overhaul in Solana’s history,” updated October 21, 2026 (vote share estimate; dashboards after activation). alchemy.com
- Solana RPC documentation, getRecentPerformanceSamples (the 60-second samples used for the TPS figures here, read from Solana’s public mainnet endpoint on October 21, 2026). solana.com
- Solana RPC documentation, getBlock (block times, parent slots and transactions for slots 459,652,999 and 459,653,008, read from Solana’s public mainnet endpoint on October 21, 2026). solana.com
- Solana RPC documentation, getAccountInfo (the Alpenglow feature account A1pengvuM6JEcyNuTnMqepBKhwHE3N6PmUrdATGawhJS, which records activation at slot 459,648,000, read from Solana’s public mainnet endpoint on October 21, 2026; the getAgGenesisCert call described on the Foundation’s Alpenglow page returned a genesis certificate for slot 459,652,999 the same day). solana.com
- ethereum.org, “The Merge” (executed September 15, 2022, completing Ethereum’s move to proof of stake). ethereum.org
More on Grey Area Labs
- LIGHT: The Solana Story, the book page
- The complete history of Solana
- Firedancer explained
- Solana vs Ethereum in 2026
- Solana DePIN in 2026
- Is Solana dead? What the record says in 2026 (publishing November 25)
- Every Solana outage, dated (publishing December 17)
Published November 16, 2026. Alpenglow’s activation was read from Solana’s public RPC on October 21, 2026 and confirmed on the Solana Foundation’s Alpenglow page on October 26, 2026; finality, TPS and validator figures are as of October 2026 and are rechecked on the publish date.