Solana's consensus rewrite is written up as arriving; its SIMD says Review
Victoria Shinder2 min readA widely shared write-up of Agave 4.3 describes Solana preparing for its largest protocol upgrade: Alpenglow, and specifically Votor, the part that replaces the existing consensus machinery of Proof of History and TowerBFT. Instead of validators voting by submitting transactions that accumulate lockout depth over 32 slots, validators exchange signed vote messages directly and the protocol aggregates those signatures into certificates.
We opened the governing proposal rather than the write-up of it. SIMD-0326, "Alpenglow", is by Quentin Kniep, Kobi Sliwinski and Roger Wattenhofer, created 25 July 2025, and it confirms the substance:
- the change is to the Votor parts of Alpenglow only;
- Rotor, the redesign of how block data propagates, is explicitly not in this proposal — the document says the network stays on Turbine for now and that Rotor "will be introduced later and will get its own SIMD";
- the two paths to finality are as described: 80% of stake notarising in one round is Fast-Finalization, 60% produces a Notarization Certificate and a second round.
Two things in the document do not match the way this is being reported.
Its status is Review, not accepted. The header field for the feature key still reads "fill in with feature key and github tracking issues once accepted". A proposal under review is not a shipping date.
The ~150 millisecond finality figure is not in the SIMD. It comes from the companion white paper and the write-ups; the proposal text does not carry it. That does not make it wrong — it makes it a design target published elsewhere, and worth attributing to where it actually appears.
⚠ The write-up that surfaced this is published by an infrastructure vendor selling Solana services, which announced an acquisition in the same banner. That is not a reason to ignore it — the technical description holds up against the SIMD — but the interest is real and the reader should have it.
What it means
The migration question for most people is smaller than the headline. If you submit transactions and read account state, there is little to change: transaction and fee mechanics stay as they are. The work lands on validators and on anything consuming blocks, votes, streams or commitment data — because vote transactions stop appearing in blocks at all, and "confirmed" and "finalized" effectively converge.
Retiring Proof of History is the part worth sitting with. It was the thing that made Solana architecturally distinctive at launch. A chain replacing its founding mechanism, with the proposal still in Review, is a story about engineering confidence as much as about latency.
The vendor write-up discussed above is at https://www.helius.dev/blog/agave-v4-3; the proposal itself is the source above.