- Anza, the firm that builds Solana’s main software, has set 21 October for the Alpenglow upgrade, targeting finality in around 150 milliseconds.
- Validator votes stop being ordinary transactions in blocks, so Solana’s reported transactions-per-second figures should fall sharply even if usage is unchanged.
- We could not find the original Anza announcement, so this is based on write-ups by CoinDesk and The Defiant, and scheduled upgrade dates have moved before.
Anza has put a date on the biggest change to Solana in years. On 21 October the network is due to switch over to Alpenglow, a rebuild of the way validators agree on blocks, with a stated target of making a transaction irreversible in roughly 150 milliseconds.
Two things change for an ordinary holder, and neither of them requires you to do anything. The first is that the number everyone quotes about Solana, its transactions per second, is about to mean something different from what it meant last month, and a big drop in it will not mean people have stopped using the network.
The second is slower to show up but matters more. How long a payment takes to become permanent decides what the network can realistically be used for. Shaving that from seconds to a fraction of one is the sort of change that an exchange, a card processor or a trading venue notices long before a price chart does.
Finality is not the same as confirmation
When you send a transaction on a , it gets picked up by the network and written into a block fairly quickly. That is confirmation, and it is what you see when a wallet says the payment has gone through.
Finality is the stronger guarantee. It is the point at which enough of the network has agreed on that block that it can no longer be undone, reordered or replaced. Until then, there is a small but real possibility that the chain settles on a different version of recent history and your transaction is no longer in it.
That gap is why exchanges make you wait before a deposit becomes usable, and why anyone building a payment product has to decide how long to hold their breath. Solana’s finality today is measured in seconds rather than fractions of one. Anza’s target of about 150 milliseconds is somewhere near the point where the wait stops being something a person would notice at all.
There is a practical reason for the push. Cointelegraph reported this week that equities, shares represented as on chains including Solana and BNB Chain, have reached close to $3.3bn in value, and that most of the trading happens outside US market hours. Round-the-clock markets are precisely the sort of thing that lives or dies on settlement speed.
Why the transaction numbers are about to fall
Here is the part that will confuse people a fortnight from now. Solana’s validators, the computers that run the network, vote on which blocks are valid. Under the current design those votes are submitted as transactions and recorded in blocks like any other, sitting alongside the swaps, transfers and mints made by actual users.
They also make up a large share of the total. By most public counts, more than half of the transactions in a Solana block have been validators voting rather than anyone doing anything.
Alpenglow takes voting out of blocks entirely and moves it into a separate consensus process. Once that happens, every chart and dashboard that counts transactions on Solana is counting a different thing from the one it counted the week before.
So expect a cliff in the figures, and expect at least some coverage to read it as collapsing demand. The useful comparison afterwards is not Solana now against Solana last month, but Solana against other networks whose numbers were never padded with votes in the first place. On that basis the drop is a cleaning of the measure rather than a loss of activity.
What still has to happen
The date is a plan, not an event that has occurred. Validators have to run new software, and the firms that supply block data to exchanges, explorers and analytics dashboards have to update too, because the shape of what they are reading changes. Upgrade dates across every major chain have slipped before when enough operators were not ready, and nothing in the reporting suggests this one is guaranteed.
We were not able to locate Anza’s own published announcement, so everything above rests on the CoinDesk and The Defiant write-ups of it. Treat the 150-millisecond figure as a design target stated by the people building it rather than a measured result.
What to watch
Whether 21 October holds, first of all. After that, watch the measured finality on mainnet rather than the target: a lab figure and a number produced by thousands of validators spread across the world are not the same thing, and the gap between them is the real test of this upgrade.
Then watch what gets built on it. Faster settlement is only worth something if payment firms and trading venues act on it, and that takes months rather than days. Our technology coverage will pick up the aftermath.