This past morning, the Solana ecosystem experienced one of the most alarming moments in its recent history. Due to a routing failure at the infrastructure operator TeraSwitch, validators controlling 28.83% of all SOL locked in staking went offline simultaneously. It was very close to the critical threshold that triggers a complete halt in transaction finalization (33.34%) — the network traveled about 86% of the way to collapse.
The incident affected approximately 90 validators. TeraSwitch's autonomous system AS20326 accounts for about 118.9 million SOL, equivalent to 27.34% of the network's total stake. At the moment of the failure, 94% of these funds went offline at once, creating unprecedented pressure on the remaining infrastructure.
Fortunately, the critical level was not reached. Routing was restored in about 33 minutes, after which the validators came back online. However, losses had already been recorded: during the downtime, network participants missed out on about 333 SOL in rewards. At first glance, the amount is modest, but the very fact of such risk concentration raises serious questions.
A Symptom of a Systemic Disease
At Marinade Finance, one of the ecosystem's key protocols, the event was described as a striking example of dangerous infrastructure centralization. The project has already announced a review of limits on the distribution of validators across autonomous systems, data centers, and backup capacities. And this is not just a formal reaction — the problem has been brewing for a long time.
Even before the incident, as of July 22, TeraSwitch served validators holding 27.1% of Solana's staked tokens. It was followed by UAB Cherry Servers (12.7%) and Latitude.sh (11%). Such concentration makes the network extremely vulnerable to single points of failure.
Some major players have already recognized the risks. For example, Coinbase reported in its first-quarter report that 13 of its validators operate through TeraSwitch, and another 10 through Latitude. At the same time, for each operator, the exchange uses a backup server in a different location to minimize the consequences of a potential failure at one provider.
Notably, this is not the first such case. In November 2022, German hosting provider Hetzner shut down servers with Solana nodes, affecting about 40% of validators with roughly 20% of the stake. The network held up then, but the current incident has shown that the safety margin is shrinking.
Additional context: in January 2026, the number of active Solana validators dropped to 800 — the lowest since 2021. At the peak in 2023, there were more than 2,500. The decline in the number of participants intensifies concentration and makes the network even more fragile in the face of infrastructure disasters.
My conclusion: Solana has once again demonstrated resilience, but this incident is a wake-up call for the entire industry. Decentralization at the consensus level is meaningless if the physical infrastructure remains in the hands of a few providers. Until the market begins to demand diversification from operators, such "near-collapses" will keep recurring.