netting

Solana activated Alpenglow on mainnet on Oct. 21 at slot 459,653,000, replacing TowerBFT with a consensus design that targets about 150 milliseconds to finality. Under the new protocol, confirmed and finalized become the same state.

At the National Securities Clearing Corporation (NSCC), multilateral netting cuts the value of equities payment obligations by an average of 98%.

Alpenglow shortens the time until a transaction becomes irreversible, and netting reduces the amount of money participants must move to complete trades.

What Solana’s Alpenglow changes

The first phase of Solana’s Alpenglow upgrade, Votor, has validators exchange votes directly and aggregate them into certificates, which allows finalization in one or two voting rounds. Transaction execution, the Solana Virtual Machine, and fee mechanics stay as they were.

The second phase, Rotor, handles block propagation and remains unscheduled.

Solana’s documentation says production applications read mostly at the confirmed level, and that every confirmed block has historically finalized.

The gain concentrates in workflows that wait for full finality before releasing assets or recognizing settlement.

Solana’s documentation cites 12.8 seconds for TowerBFT finality at the older 400-millisecond slot time. Solana cut slots to 200 milliseconds on Oct. 9. Applying TowerBFT’s documented 32-slot finalization rule to that slot time gives a nominal 6.4-second baseline, rather than a measured end-to-end settlement time.

The 98% figure

NSCC acts as central counterparty for eligible equities trades, guaranteeing qualifying obligations and offsetting purchases and sales among participants so residual payments settle.

The Depository Trust & Clearing Corporation (DTCC) reports the 98% average reduction in payment value and $608 trillion of equity transactions processed in 2025.

The figure covers payment obligations and is separate from margin, regulatory capital, and peak intraday funding. It also reflects a market structure with central clearing, participant agreements, and risk management.

NSCC settles on T+1, the gap between trade date and settlement date, which differs from the time a system takes to process a payment.

Solana’s delivery-versus-payment program shows that both parties fund separate escrow accounts, a settlement authority signs the exchange, and both legs execute atomically.

Applications are told to recognize settlement when the transaction is finalized, and the documentation separates network finality from finality under the parties’ legal arrangements. Faster finality shortens the last step, and the parties must source and commit the assets beforehand.

The capital-reuse math

A sensitivity model assumes strictly sequential settlements, identical transactions, and instant reuse of funds, isolating latency as a measure separate from institutional savings.

From the nominal 6.4-second baseline to 0.15 seconds, the theoretical turnover multiplier is 42.7 times when consensus time is the full cycle and 10.6 times with half a second of other processing. It falls to 3.9 times with two seconds and 2.2 times with five, and to about 1.1 times with 60 seconds.

Against the 12.8-second baseline, the same steps give 85.3, 20.5, 6.9, 3.5 and 1.2 times.

Custody checks, funding transfers, compliance reviews and signatures are the usual sources of non-consensus delay, so the value of faster consensus depends on what holds capital back.

A turnover multiplier measures how often funds can cycle through serial settlements. It is separate from reductions in broker collateral or clearinghouse margin.