What Exactly Happened?
On the morning of August 12, 2026, the Solana network came close to the point beyond which it would stop finalizing new blocks. According to the staking protocol Marinade Finance, validators responsible for 28.83% of staked SOL stopped voting, a condition known as delinquency (a validator is online but unable to vote, or entirely offline).
The critical threshold at which the network loses finality sits at 33.34% of the staked share. According to Marinade, Solana got roughly 4.5 percentage points below that line, meaning about 86% of the way to a finalization halt (a figure cited by both Incrypted and Decrypt).
Finality means a block is considered irreversibly confirmed. When too large a portion of validators stops voting, the network can still produce blocks, but it cannot close them with certainty.
What Was the Cause?
According to Decrypt, the outage was caused by a malformed default route at a single hosting provider, the company TeraSwitch. According to Incrypted, validators with more than 27% of total stake run on this provider's infrastructure, so a single network error at one provider was enough to knock this large a share of the voting power offline.
This is a textbook example of concentration risk: technically it was not a flaw in the Solana protocol nor an attack, but an outage at a single infrastructure supplier that a large group of validators depends on.
How Long Did It Last and What Was the Damage?
| Metric | Value | Source |
|---|---|---|
| Share of stake in delinquency | 28.83% | Marinade Finance / The Defiant |
| Critical finality threshold | 33.34% | Incrypted |
| Progress toward loss of finality | approx. 86% | Incrypted, Decrypt |
| Duration | approx. 33 minutes | Incrypted |
| Validators affected | 90 | Marinade Finance / The Defiant |
| Lost rewards | 333 SOL total | Marinade Finance / The Defiant |
According to Incrypted, the problem was resolved after roughly 33 minutes. Marinade states that the incident affected 90 validators and cost them a total of 333 SOL in rewards over that period.
What Does This Mean for Network Resilience?
Solana did not lose finality in this case. The data, however, shows how thin the margin was: had the outage hit a few percentage points more of the stake, the network would have stopped finalizing blocks. This is a concrete illustration of why distributing validators across different infrastructure providers (and different geographic locations) is a technical parameter, not just a cosmetic one.
Alongside this incident, CoinDesk reported on August 12 that the Solana-built network DoubleZero is adding the order book of the prediction market Kalshi to its low-latency data feed, a move aimed at institutional demand. No direct link to the TeraSwitch outage emerges from the available sources; we mention it only as context for how additional infrastructure layers are being built around Solana in parallel.
What Remains Unclear?
The verified sources do not include an official statement from TeraSwitch itself nor the precise technical cause of the faulty route beyond the description of a "malformed default route." Nor do we have information on whether and how validator operators plan to reduce their dependence on a single provider. These are the points worth watching in incidents of this type: a post-mortem from the provider and any changes in stake distribution.