On August 12, 2026, a routing failure at infrastructure provider TeraSwitch pushed Solana to the brink of a transaction finality halt, temporarily sending 28.83% of all staked SOL into delinquency. The network continued producing blocks throughout the incident, which came within roughly four and a half percentage points of the 33.34% threshold at which Solana would cease finalizing transactions. Solana, like other Proof-of-Stake networks using Byzantine Fault Tolerance, requires at least two-thirds of active stake to participate in voting to finalize blocks—meaning a single infrastructure failure cascading across enough validators could theoretically halt consensus even without a protocol-level bug.

The disruption originated when a default route advertised from TeraSwitch's Miami site propagated across the provider's backbone, triggering simultaneous disconnections at twelve locations spanning Europe and Asia-Pacific. Marinade Finance, the liquid staking protocol that first identified the outage, reported that approximately 90 validators were affected.

"Solana got 86% of the way to a halt this morning and it barely registered anywhere," Marinade wrote on X. "28.83% of staked SOL went delinquent. Finality stops at 33.34%."

The Solana Foundation stressed that at no point did the network stop processing transactions. Jacob Creech, a technology executive at the Foundation, stated:

"Last night an infrastructure provider used by some Solana validators had a failure. You probably didn't notice, because the network didn't: blocks kept producing and transactions kept landing."

Last night an infrastructure provider used by some Solana validators had a failure. You probably didn't notice, because the network didn't: blocks kept producing and transactions kept landing. The facts: – The Solana network remained operational – 597 of 699 staked validators… — Jacob Creech (@jacobvcreech) August 12, 2026

Of 699 staked validators, 597—approximately 85%—continued voting during the incident. Validators enrolled in the Solana Foundation Delegation Program, which delegates SOL to validators meeting specific operational and geographic distribution criteria, were unaffected. Affected nodes recovered within 33 to 40 minutes.

TeraSwitch's autonomous system, AS20326, held 118.9 million SOL at the time of the incident, representing more than a quarter of the network's total stake. According to Marinade, 94% of that stake went dark simultaneously. TeraSwitch restored routing after approximately ten minutes of diagnosis and 33 minutes of total downtime, though its Miami site remained disconnected from the backbone pending further investigation.

Infrastructure Concentration and Failover Gaps Surface

While Solana's consensus layer withstood the disruption, the outage brought infrastructure concentration risk into sharp focus. The incident illustrates a well-documented tension in PoS networks: although stake can be distributed across many validators on-chain, those validators frequently run on shared physical infrastructure, creating hidden correlations that governance metrics may not capture. Marinade's post-incident analysis revealed that the disruption extended beyond TeraSwitch: an additional 14.1 million SOL dropped offline concurrently across Latitude.sh, Limestone, Butterfly Research, and Allnodes—pointing to either a shared dependency or an unlikely coincidence.

"We can't tell from the data why," Marinade stated. "Either way, counting stake by hosting provider undercounts what actually falls over together."

6/ It didn't stop at that ASN too. Another 14.1M SOL went down in the same minutes on @latitudesh , Limestone, Butterfly Research and @Allnodes . We can't tell from the data why. Some shared dependency, or coincidence. Either way, counting stake by hosting provider undercounts… — Marinade (@MarinadeFinance) August 12, 2026

Recovery data further underscored redundancy shortcomings. Fifty-nine validators representing 80.2 million SOL came back online within the same narrow window across Amsterdam, Frankfurt, and Tokyo, indicating that automated failover mechanisms did not activate.

"Nobody failed over," Marinade observed. "They sat there until the routing reconverged."

Helius, Solana's second-largest validator, remained offline for the entire 33-minute duration. Of the 74 validators Marinade was able to monitor, only three—operated by Sol Strategies and Lion3d—recovered cleanly.

Calls for Stricter Distribution Standards

The incident has prompted calls within the ecosystem for tighter distribution requirements. Marinade announced plans to review concentration caps across autonomous systems, data centers, and backup infrastructure—an effort that could set a precedent for how liquid staking protocols assess and enforce decentralization beyond on-paper validator counts.

Some operators already employ multi-provider strategies. Coinbase disclosed in its first-quarter report that it distributes its validators across TeraSwitch and Latitude.sh, with backup servers in separate locations to mitigate single-provider exposure.

The outage also drew comparisons to November 2022, when German hosting provider Hetzner disconnected servers running roughly 40% of Solana validators—again without halting the network. The latest disruption, however, arrives amid a broader decline in validator participation: active Solana validators fell to 800 in January 2026, the lowest figure since 2021 and down from a peak above 2,500 in 2023.

With stake increasingly concentrated among fewer operators and shared infrastructure layers, the episode highlights how protocol-level decentralization can be undermined by physical-layer dependencies. How the ecosystem responds—whether through voluntary standards, protocol-level incentives, or staking-protocol-driven mandates—may shape resilience expectations not only for Solana but for PoS networks generally.