US spot Solana ETFs lost a net $71.47 million in the week of 26 to 30 October 2026, the largest weekly outflow since the funds launched on SoSoValue's data, and the fourth outflow week in a row. It landed days after Alpenglow, Solana's first consensus replacement since 2020, went live on mainnet on 21 October. SOL has fallen about 15% since the 21 October close, to $91.21 (CoinGecko, 2 November, 19:59 UTC). The upgrade is real, Anza's claimed 150 millisecond finality has no named third-party confirmation yet, and nothing in the data ties the outflow to the upgrade.
Key Takeaways
The $71.47M outflow is a record for these funds, but cumulative net inflows are still $1.47B and net assets are $1.40B, roughly 2.5% of SOL's market cap. Bitwise's BSOL accounts for roughly 64% of the outflow. Alpenglow is live (slot 459,653,000, 21 October). Anza claims about 150 ms finality; the only third-party sample found, by an unnamed party timing a public RPC node, measured a 404 ms median over 24 hours. That is a different measurement from Anza's, so the two numbers are not a direct contradiction. The outflow and the upgrade are two separate stories that happened in the same stretch of days.
What Happened to Solana ETF Flows
Net outflows from US spot Solana ETFs totalled $71.4748 million between 26 and 30 October, per SoSoValue data reported by PANews and TokenPost on 1 and 2 November. Both outlets call it the largest weekly withdrawal since the products launched in October 2025. FXStreet, also citing SoSoValue, adds that it was the fourth straight week of outflows, a run that began in early October.
Aggregate net assets across the funds stand at $1.40 billion, which SoSoValue puts at 2.49% of SOL's market capitalisation at its own snapshot (our CoinGecko figure of $53.83B would give about 2.6%). Cumulative net inflows since launch are $1.47 billion. Net assets are below cumulative inflows partly because the price of SOL has fallen, not only because of redemptions. The week's outflow is roughly 5% of current net assets.
| Fund | Sponsor | Net flow, 26 to 30 Oct | Cumulative net inflows |
|---|---|---|---|
| BSOL | Bitwise | -$45.82M | $1.12B |
| GSOL | Grayscale | -$25.71M | $128M |
| MSOL | Morgan Stanley | +$49,300 | $43.11M |
| Other funds | Not itemised by the sources | none implied (the three funds sum to the total) | about $179M (residual, our arithmetic) |
| All funds | -$71.47M | $1.47B |
Who Sold: One Fund Carries Most of the Outflow
Bitwise's BSOL lost $45.82 million, about 64% of the total, and has $1.12 billion of cumulative net inflows. Grayscale's GSOL lost $25.71 million and has $128 million of cumulative net inflows. Morgan Stanley's MSOL took in $49,300, the only inflow of the week, and has $43.11 million cumulative. The concentration matters: a single large holder or one allocator rebalancing out of BSOL would look like a market-wide exit in the headline number.
BSOL's history makes the reversal notable. In late 2025 it logged 33 straight days of inflows. The same fund now leads the redemptions. Flow data cannot show whether the holders redeeming now are the same ones who bought during that streak, or why they are selling.
What the Record Does and Does Not Mean
A record weekly outflow is a statement about flow, not about stock. With $1.47 billion of cumulative inflows still on the books, the funds remain net positive since launch. The label "all-time high" also rests on a single data provider: SoSoValue is the source for every article we found, and neither PANews nor TokenPost shows a prior-week comparison in the text. We could not reach Farside, the other common tracker, to cross-check it (the site returned a 403 to our request). PANews, TokenPost and FXStreet agree, but they are one dataset repeated, not three independent measurements.
Where SOL Trades After Five Weekly Losses
SOL was at $91.21 at 19:59 UTC on 2 November, down 2.84% over 24 hours, with a market capitalisation of about $53.83 billion and 24-hour volume of about $2.50 billion (CoinGecko API). That is roughly 15% below the $107.39 close of 21 October, the day Alpenglow activated.
FXStreet's technical read on the morning of 2 November put SOL below all three major moving averages: the 50-day EMA at $103.81, the 100-day at $99.96 and the 200-day at $98.08. It also counted five consecutive weekly losses since late September, an RSI near 31 (close to oversold but not decisive) and a negative MACD. Its listed support levels are a broken rising trendline near $88.42, then horizontal support near $78.35. SOL has since moved lower, so the first of those is about $3 away. These are one analyst's chart levels, not forecasts.
The Timing of the Decline
Our own daily series shows the price drifting between about $100 and $107 from 21 to 28 October, then dropping from $102.38 on 28 October to $96.88 on 29 October, a one-day fall of about 5.4%. The ETF outflow week (26 to 30 October) overlaps that drop. Overlap is not cause: the data we have does not separate outflows driving the price from the price falling and prompting redemptions, and we found no verified explanation for the 29 October move. We would treat any confident explanation with caution.
What Alpenglow Changed on 21 October
Alpenglow reached mainnet-beta at slot 459,653,000 (epoch 1064), shortly before 14:00 UTC on 21 October 2026, replacing TowerBFT, the voting system Solana had used since its March 2020 launch (The Block, Solana Compass). Anza's announcement is timestamped 13:54:56 UTC, and the migration waited 5,000 slots after the epoch boundary, roughly 17 to 18 minutes (Solana Compass). Anza says the upgrade went through public testing that included red teaming, penetration testing, repeated migration rehearsals and a security competition (The Block).
How Votor Finalizes Blocks
The first phase is called Votor. Validators send votes directly to each other and bundle them into certificates. A block finalizes in one round when validators holding 80% of stake vote for it, or in two rounds with separate 60% certificates. Condensing Solana's old two-step confirmation into one step is where the finality gain is supposed to come from (The Block, 21 October).
Votes Left the Blocks
Vote transactions are no longer included in blocks. Solana Compass reports that votes made up about 62% of Solana's processed transactions in the week before the switch, so any transaction count from 21 October onward covers user activity only and cannot be compared with earlier totals. A dashboard showing "transactions down 60%" after the upgrade is measuring a change in what is counted, not a drop in usage.
What It Costs Validators
Per-vote fees end. Each voting validator now pays a flat 0.8 SOL per epoch through a Validator Admission Ticket, deducted from the vote account. The first ticket was charged at the start of epoch 1064 and covers epoch 1065. At the current price that is about $73 per epoch per voting validator (0.8 SOL x $91.21, our arithmetic). The epoch length is not given in our sources, so we do not annualise it. Operators in the active set had to run version 4.3.2 of an Agave-based client (Agave, Jito-Solana or Harmonic Salsa).
Firedancer and Frankendancer Were Not Supported at the Switch
Firedancer and Frankendancer were not supported during the switch. Firedancer 26.10 supports Alpenglow, so operators can return to it after the transition. Frankendancer will not be supported once Alpenglow is active. Readiness was a concern beforehand: Solana Compass cites a 19 September readiness tracker that flagged 45% of stake as not ready, and says the completed handoff confirms only that validators with at least 82% of stake took part. No count of ready stake at the switch has been published.
What Did Not Change
Apps that send and sign transactions need no changes, because the SVM, transactions, programs and fees are unchanged. Services that read blocks in real time must track a new bank_id field, since a slot can now carry more than one candidate block. Rotor, the planned replacement for block propagation, is "not yet scheduled". Geo-aware leader rotation (SIMD-0674 and SIMD-0675) and Quantumglow, a post-quantum variant, have no committed dates.
Did Finality Really Fall to 150 Milliseconds?
Not yet confirmed. Anza claims finality fell from about 12.8 seconds to about 150 milliseconds. That is Anza's figure, not a third-party measurement. The Solana Foundation says actual performance will depend on network conditions (Solana Compass), and Anza's chief economist expects users to see 150 to 200 ms (Solana Compass). The 12.8 second baseline reflects older 400 ms slots, and Galaxy estimates TowerBFT finality at about 6 seconds with today's 200 ms slots (as cited by Solana Compass), so the improvement is smaller against that baseline, though still large.
| Figure | Value | Who says so | Caveat |
|---|---|---|---|
| TowerBFT finality, headline baseline | about 12.8 s | Anza | Reflects older 400 ms slots |
| TowerBFT finality at today's 200 ms slots | about 6 s | Galaxy (estimate) | Estimate, not a measurement |
| Pre-upgrade median, public sample | about 7 s | TokenPost | Whole-second block times, a different method from the post-upgrade sample |
| Alpenglow, Anza's claim | about 150 ms | Anza | Not yet confirmed by a named third party; Anza's chief economist expects 150 to 200 ms for users |
| Alpenglow median, 24 h sample starting about 22 h after activation | 404 ms | TokenPost | Public RPC observation, 357,000+ slots, 24 h to 23 Oct 12:30 UTC |
| Alpenglow 90th / 99th percentile | 579 ms / 765 ms | TokenPost | Same sample |
The one third-party sample we found is a TokenPost report published 27 October. It timed how long a public RPC node took to report a root at or above a slot after first reporting that slot, across more than 357,000 slots in the 24 hours to 23 October 12:30 UTC. The median was 404 ms, the 90th percentile 579 ms and the 99th percentile 765 ms. TokenPost does not name who ran the measurement and warns that the pre-upgrade comparison uses a different method (whole-second block times against millisecond WebSocket observations), so the improvement factor cannot be read from it. It measures how fast a public node reports a root, which is not the same thing as Anza's finality figure. It also does not say how much of the gap between 404 ms and 150 ms is RPC reporting delay rather than consensus. We found no multi-week, same-method measurement.
"Alpenglow is live on mainnet-beta. Solana's new consensus protocol: finality reduced from 12.8 seconds to 150 milliseconds, built by Anza." (Anza on X, 21 October 2026; emoji omitted)
Does a Faster Chain Support the Price?
This section is our reading, not reported fact. Faster finality helps applications where waiting matters: trading venues, payments and anything that must know a transaction is final before the next step. It does not change how much SOL is issued, how many people hold it or what an ETF allocator pays for it. The price and flow data above show no sign that the upgrade moved demand, and SOL fell about 15% in the twelve days after it. That does not show the upgrade failed. A technical gain that was widely expected can be priced in before it ships, and a market short of buyers can overwhelm a good upgrade. The data cannot tell those apart.
The ETF side tells a related story. Our earlier coverage of the governance vote on SGP-0002 and SGP-0003 explains the changes to SOL issuance and staking yield that went to a vote in August. In our reading, flows tend to follow price and risk appetite more than protocol design, though this data cannot test that. A week of outflows during a five-week price slide is the pattern you would expect from price-sensitive holders, and equally consistent with a handful of large rebalancing trades. We cannot distinguish them from public data.
What to Watch Next
- Weekly ETF flows. A fifth outflow week would extend the run. A flat or positive week would put the record in context as a single washout.
- The $88.42 and $78.35 levels. These are the supports FXStreet lists below the current price.
- A same-method finality measurement. The useful question is whether the median settles near 150 ms or stays near 400 ms across weeks, with the measurer named.
- Rotor scheduling. It is the second phase and has no date.
- Firedancer's return. Version 26.10 supports Alpenglow, so the share of stake running it is worth tracking.
Known Discrepancies
Several numbers in circulation do not match each other or the sources behind them. We list them so you do not have to untangle them yourself.
| Claim in circulation | What the sources support |
|---|---|
| Alpenglow cut finality to 150 ms | 150 ms is Anza's figure. The only third-party sample found (unnamed party, public RPC node) measured a 404 ms median, a different measurement. |
| Finality fell from 12.8 s | 12.8 s is Anza's baseline. Galaxy estimates about 6 s at current slot times, and a public sample used about 7 s. |
| SOL trades near $95 | $95.48 was a Monday morning GMT reading (FXStreet). CoinGecko showed $91.21 at 19:59 UTC the same day. Different timestamps, not a different fact. |
| A record weekly outflow | True in SoSoValue's dataset, which every outlet repeats. Only one data provider was found, so the record rests on one dataset. |
| Transactions fell after 21 Oct | Vote transactions (about 62% of processed transactions the week before) left the blocks. Post-switch totals count user activity only and are not comparable. |
Sources and Methodology
We took the weekly ETF figures from SoSoValue data as reported by PANews (2 November) and TokenPost (1 to 2 November), and the "fourth straight week" and technical levels from FXStreet (2 November, 05:38 GMT). The price series and live quote come from the CoinGecko API, queried on 2 November 2026; the live reading is timestamped 19:59:17 UTC and the daily series uses the last data point of each UTC day. Alpenglow details come from The Block and Solana Compass (21 October) and Anza's announcement. The finality sample comes from TokenPost (27 October). The 6 second TowerBFT estimate is Galaxy's, as cited by Solana Compass. Where outlets disagreed we cite both and say what each measures.
Items we looked for and could not verify, so left out: a reported Linux kernel advisory for validators, a fake "SKR Claim" app on Solana Mobile, and a merged p-ATA proposal reported without a primary source. Farside returned a 403, so the ETF record has not been cross-checked against a second data provider.
- PANews, "Spot SOL ETFs Post Record $71.47M Weekly Net Outflows", 2 November 2026
- TokenPost, "Solana Spot ETFs Set Weekly Outflow Record at $71.47 Million", 1 November 2026
- FXStreet, Solana price forecast, 2 November 2026
- CoinGecko API, SOL price and market data, 2 November 2026
- The Block, "Anza's Alpenglow goes live on Solana, cutting finality to 150 milliseconds", 21 October 2026
- Solana Compass, "Solana Alpenglow Live on Mainnet: TowerBFT Retired, Anza's 150ms Finality and What Changes", 21 October 2026
- TokenPost, "Solana's Alpenglow Logs 404-Millisecond Median Finality in 24-Hour Sample", 27 October 2026
- Anza on X, Alpenglow activation announcement, 21 October 2026
This article is analysis, not investment advice. Prices and flows change quickly, and every figure here carries the date it was checked.