Why
In 2026 a routing error on Solana took roughly 29% of stake offline at once and the network came within a few percentage points of losing finality. Nothing was hacked and no key was stolen — the consensus math worked exactly as designed. The failure was that a single infrastructure fault could reach a third of the validator set at the same time, which is a question about network topology, not cryptography. The interesting output is a number nobody publishes: for a given chain, what is the largest slice of stake that shares one correlated point of failure, and how does that compare to the fraction that halts finality (33% for both Solana's and Ethereum's BFT thresholds)?
How it works
Planned as a measurement, not an essay: pull the active validator set and its stake weights from public RPC, resolve each advertised gossip/TPU endpoint to an IP, map those to ASN and hosting provider with a public IP-intelligence dataset, then aggregate stake by ASN, by provider, and by geographic region. The headline figure is the stake-weighted Herfindahl index plus the single largest correlated bucket, both plotted against the 33% halt line. The same pipeline runs unchanged against Ethereum's beacon chain for comparison, where the concentration hides one layer down — in staking pools and the handful of clouds their operators rent from rather than in the validator count itself.