Solana came within roughly 20 million $SOL of losing transaction finality on Wednesday after a routing failure at infrastructure provider Teraswitch caused 28.83% of the network’s staked $SOL to become delinquent.
Solana stops finalizing transactions if more than 33.34% of its staked $SOL goes offline. Marinade Finance calculated that the incident pushed the network to 86% of that threshold.
The mainnet did not halt. Blocks continued to be produced, and transactions continued to land. However, the incident exposed how a failure at a single infrastructure provider can affect a large portion of the network at once.
The affected validators lost about 333 $SOL in rewards, which validator bonds will cover at the end of the epoch.
Routing Fault Spread Across Europe and Asia
The incident began when Teraswitch advertised a default route from its Miami facility without the expected attributes. A route reflector in Amsterdam then distributed that route to sites across Europe and Asia.
Edge routers at those locations treated the route as their local default and preferred it over the correct route. Core infrastructure rejected the route as invalid, leaving 12 sites without a valid forwarding path.
The affected locations included London, Amsterdam, Dublin, Frankfurt, Singapore and Tokyo. North America did not experience the same disruption.
Teraswitch identified the issue within about 10 minutes, and traffic returned by 04:16:15 UTC. Miami remained off the provider's backbone while Teraswitch investigated the source of the route advertisement.
Around 90 Solana validators experienced the outage for roughly 33 minutes.
Stake concentration amplified the impact
Marinade's analysis showed that AS20326 held 118.89 million $SOL, more than one-quarter of all staked SOL. About 94% of that stake went offline simultaneously during the incident.
Another 14.1 million $SOL also went offline across Latitude.sh, Limestone, Butterfly Research, and Allnodes. Marinade said the available data could not establish whether those outages resulted from a shared dependency or coincidence.
That distinction matters because measuring concentration by hosting provider alone may underestimate how much stake can fail together.
Marinade also found that 59 validators holding 80.2 million $SOL returned within the same narrow recovery window across Amsterdam, Frankfurt, and Tokyo. Those validators did not appear to trigger automatic failover. They remained offline until routing reconverged.
Helius, Solana's second-largest validator, stayed offline for the full 33 minutes. Of 74 validators Marinade could measure, only 3 returned cleanly through their redundancy arrangements: Laine, Cogent Crypto (both operated by Sol Strategies), and Lion3d.
Marinade also acknowledged concentration within its own delegation program. Four ASNs account for two-thirds of the stake it allocates, while AS395201 alone represents 36.94%. The organization said it plans to review concentration limits by ASN and data center and publish whether validators use hot-swap and automatic failover systems.
Solanabeach data currently puts Solana's superminority at 17, underscoring the relatively small number of validators that collectively hold enough stake to influence network finality.
Solana's Mainnet Kept Running
Solana Foundation’s Vice President of Technology, Jacob Creech, emphasized that 597 of 699 staked validators continued voting, or roughly 6 out of 7. Affected validators recovered within 40 minutes, while validators in the Solana Foundation Delegation Program remained unaffected.
Anza's trent.sol added another detail: Solana's Devnet actually halted during the same Teraswitch issue but resumed automatically once the routing problem cleared. The Devnet and mainnet run the same software, suggesting that Solana's software has gained resilience compared with earlier incidents.
Anza performance team lead Alessandro Decina opined that a similar event would have brought Solana down 2 years ago, describing the network's progress as a result of software improvements.
The February 2024 Outage Remains the Warning
Solana last halted outright in February 2024, and the network took nearly 5 hours to restart. Since then, Solana has maintained 100% uptime for 30 consecutive months without network downtime.
The latest incident therefore presents a mixed picture. Solana demonstrated that its mainnet could continue operating despite almost 29% of its stake going offline. At the same time, the event showed that infrastructure concentration can bring the network uncomfortably close to its finality threshold.
Solana co-founder Greg Fitzgerald called the incident an uncomfortable reminder that decentralization remains one of crypto's hardest problems.
The Outage Reached Beyond Solana
Pyth also experienced disruption because Teraswitch hosted 2 of its 4 NATS nodes. When those nodes became unreachable, NATS lost quorum, and dependent Router services became unavailable.
Pyth said it had already begun reducing infrastructure concentration and will accelerate that work by adding another infrastructure provider and distributing critical services across independent failure domains.
The incident ultimately left Solana's mainnet running, but it also demonstrated how quickly infrastructure dependencies can turn a localized routing failure into a network-wide risk. If delinquent stake had crossed 33.34%, Solana would have stopped finalizing transactions for everyone, regardless of where they held their $SOL.
Read More on SolanaFloor
Pump.fun vs. Fomo Gets Serious as Traders Question Fomo’s Fees
SEC to Roll Out “Major Initiatives” to Turbocharge Crypto Industry as CLARITY Flounders
Solana Wants To Burn 14x More $SOL
