A routing failure knocked roughly 29% of staked SOL offline, pushing Solana within about 4.5 percentage points of the threshold that would freeze transaction finality. Validators that went dark lost staking rewards but faced no slashing, reviving debate over the network's resilience and its planned Alpenglow upgrade.
Solana came close to a finality halt after a routing failure took roughly 28.83% of staked SOL offline, leaving the network about 4.5 percentage points from the point where it can no longer confirm transactions as permanent.
How close Solana came to a halt
Solana's consensus mechanism, Tower Byzantine Fault Tolerance, needs roughly two-thirds of all staked SOL, about 66.67%, actively participating to finalize transactions. As a result, once more than 33.34% of stake goes dark, the network loses the supermajority it needs, and blocks can still be produced but nothing gets stamped as irreversible. Validators that go offline on Solana face no slashing penalty; instead, they simply stop earning rewards.
A streak tested for the first time since 2024
Before the incident, Solana's official status page had shown all systems operational for over 30 months straight, with the network's last full outage dating back to February 2024. However, a network can keep producing blocks while losing the ability to finalize them if enough stake goes delinquent, so the routing failure exposed a scenario where the chain stayed technically "up" while its finality guarantee came under strain. Reports from 2026 have shown up to 32 validator delinquencies within a 30-day window on Solana, but nearly 29% of stake going down at once points to a systemic issue rather than scattered individual failures.
Alpenglow's safety-first design
Solana has been developing a protocol upgrade called Alpenglow, which aims to compress transaction finality to approximately 100-150 milliseconds. The upgrade prioritizes safety over liveness, so it is built to pause block production entirely rather than risk confirming transactions that might later prove inconsistent. It also introduces a fault-tolerance model that distinguishes actively malicious validators from those simply offline due to passive failures like the routing issue behind this incident.
The absence of slashing is likely to draw renewed debate: proponents say it keeps the validator set accessible, while critics counter that without financial consequences for downtime, validators have less incentive to build the redundancy that prevents large-scale simultaneous failures. Solana's consensus model tolerates up to a third of stake going offline, and this incident tested that boundary more aggressively than anything since the February 2024 outage.
Source: Crypto Briefing
Trading involves risk.