A routing failure on the Solana network pushed 28.83% of staked SOL offline for approximately 33 minutes, affecting roughly 90 validators. Despite losing nearly a third of participating stakes, the network continued finalizing transactions, staying just 4.51 percentage points above the critical 33.34% finality threshold. Validator diversity and rapid recovery prevented a broader disruption, though the outage cost validators around 333 SOL in rewards and exposed correlated infrastructure risks.
A routing failure exposed how quickly infrastructure issues can spread across validators sharing common dependencies on the Solana network. The disruption pushed 28.83% of staked SOL offline, affecting roughly 90 validators for about 33 minutes.
This left only 4.51 percentage points, or roughly 19.9 million SOL, before hitting the crucial 33.34% finality threshold. Despite this setback, transactions continued finalizing, showing consensus remained functional despite losing nearly a third of participating stakes.
As routing recovered, delinquent stakes gradually fell toward zero, preventing broader disruption for users. However, validators lost around 333 SOL in rewards, creating a direct economic cost from the outage.
That infrastructure weakness ultimately tested whether Solana could retain enough voting power to keep confirming transactions. Despite 28.83% of the stake going offline, another 71.17% remained active, keeping participation above the two-thirds requirement.
Transaction finalization did occur, which prevented users from experiencing delays on their pending transactions. The probability of a second correlated outage occurring simultaneously increased significantly.
The ability to survive this outage was in part due to validation services outside the impacted routing group continuing to function normally. Although AS20326 had more than 25% of the total SOL staking pool online at the time, this accounted for approximately 94% of their stake failing.
Validators were distributed across many providers and geographies, maintaining connectivity and preventing the failure from cascading. At the time of the outage, 699 validators had stakes on Solana, and 597 of them—85%—continued to vote for blocks.
Recovery strengthened this buffer as affected validators gradually rejoined within roughly 40 minutes. This restored voting participation before another infrastructure failure could compound the disruption.
Geographic and provider diversity limited the outage, while rapid recovery prevented concentrated infrastructure exposure from becoming systemic. Another correlated outage affecting additional stakes could cross the 33.34% threshold, turning infrastructure disruption into a network-wide halt.
