Routing Fault at Teraswitch Took 28.83% of Solana’s Stake Offline, Marinade Says

Solana came close to losing its ability to finalize transactions early Wednesday after a routing fault at hosting provider Teraswitch disconnected 28.83% of staked SOL, according to staking platform Marinade. The outage stopped short of the network’s 33.34% finality threshold, but the incident highlighted how concentrated infrastructure can threaten a distributed validator set.

How the Routing Failure Spread

The fault began when a default route from Teraswitch’s Miami facility was stripped or misconfigured and then propagated across sites in Europe and the Asia-Pacific region. Marinade said the disruption placed the network approximately 86% of the way to a freeze, based on the amount of stake that went offline.

The largest concentration was tied to autonomous system AS20326, which carried 118.89 million SOL—more than one-quarter of all staked SOL. Marinade reported that 94% of the stake associated with that system disappeared during the same period. A further 14.1 million SOL went offline across latitude.sh, Limestone, Butterfly Research and Allnodes, though Marinade said its data could not explain the cause of those additional outages.

The incident also raised questions about the effectiveness of validator failover. Marinade found that 59 validators holding 80.2 million SOL returned within a narrow window in Amsterdam, Frankfurt and Tokyo after waiting for routing to reconverge rather than switching to alternative infrastructure. Helius, Solana’s second-largest validator, remained offline for the full 33 minutes measured by Marinade.

Recovery and Competing Assessments

Of the 74 operators Marinade could measure, only three recovered cleanly: Laine, Cogent Crypto—both operated by Sol Strategies—and Lion3d. The affected validators lost 333 SOL in rewards, a shortfall that validator bonds are expected to cover at the end of the epoch.

The Solana Foundation offered a more restrained assessment. Vice President of Technology Jacob Creech said the network continued producing blocks, with 597 of 699 staked validators still voting. He added that affected validators recovered within 40 minutes and that validators participating in the Foundation’s delegation program were unaffected, describing the event as evidence that infrastructure diversity was working.

Marinade’s analysis also identified concentration within its own allocation model:

• Four autonomous systems hold roughly two-thirds of the stake it distributes.
• One system accounts for 36.94% of that allocation.
• The affected event pushed one external system above the 25% ceiling used by the Foundation’s delegation program.

Marinade said it will review concentration limits by network and data center and begin publishing which validators use hot-swap systems and automatic failover.

What This Means

The episode was not an outright Solana halt, but it exposed a narrow margin between continued block production and a loss of finality. The network’s last outright halt, in February 2024, took approximately five hours to restart; this event recovered far more quickly, yet the scale of stake tied to one hosting route remains a material resilience risk.

For SOL stakeholders, the immediate impact was limited reward loss rather than a prolonged service interruption. Longer term, operators may face greater pressure to diversify across autonomous systems, regions and data centers, while delegators could increasingly treat automated failover and infrastructure independence as core criteria for allocating stake.