Morning Edition · Wednesday, September 16, 2026Published at 1:44 AM EDT · New York
Telemetry reviewed after the incident shows the network reached about 86 percent of the level at which block finality stops, and a single provider hosts more than a quarter of all staked SOL.

A newly detailed account of a routing failure at hosting provider Teraswitch shows how close Solana came to losing block finality. Roughly 29 percent of staked SOL went offline, within about 4.5 percentage points of the 33.34 percent at which the network stops finalizing blocks. CryptoSlate reported that the failure stayed below the halt threshold while independent telemetry showed delayed finality and sharply lower throughput.
The cause was a network-routing mistake, not a consensus bug and not an attack. According to an analysis by the staking protocol Marinade, Teraswitch advertised a default route from its Miami site without the attributes that describe it. A route reflector in Amsterdam then propagated that route across sites in Europe and Asia, and edge routers preferred it over the valid one. Teraswitch's own core then rejected the route as invalid, leaving twelve sites with no usable path. About 90 validators and more than 133 million SOL dropped out at once. Engineers found the software bug in roughly ten minutes, and full recovery took about 33 minutes.
The concentration number is the part that should concern stakers. Marinade's review found that a single provider hosts 27.34 percent of all staked SOL, above the Solana Foundation's 25 percent guideline, and that 94 percent of the stake on that provider went offline at the same time. Validator rewards lost during the episode were small, around 333 SOL, but the risk being measured is not lost rewards. It is whether a chain that reports thousands of validators actually has a few dozen correlated failure points.
Solana holds $5.73 billion in decentralized finance (DeFi) total value locked according to DefiLlama, second among chains behind Ethereum's $48.82 billion. Applications settling that value inherit the hosting concentration of the validator set, whether or not their users are aware of it.
Start a discussion in Townsquare.
More from this edition
Marinade, which published the analysis, benefits from a concentration argument that supports redirecting delegation, and competing proof-of-stake networks benefit from a Solana-specific fragility narrative.
The routing sequence and the 4.51-point margin come from Marinade's review and are corroborated by CoinDesk, but the incident occurred on 12 August 2026 rather than this week, the stake figure is 28.83 percent rather than 29, and Marinade states finality was never actually interrupted, which sits against the article's description of delayed finality.
An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.
What this means
Stake counts and validator counts describe decentralization badly when many validators share one data-center operator and one routing path, so a vendor configuration error becomes a consensus-level event. Holders of assets on high-throughput chains and the lending and perpetuals venues built on them carry that correlation risk directly, because a finality halt freezes liquidations and oracle updates at the same time. The fix is boring and measurable: stake pools and foundations either enforce hosting diversity limits with real penalties, or the next routing error lands above the threshold instead of below it.
What to watch
Observations to monitor, not financial advice.
Synthesized from: CryptoSlate · SolanaFloor · Metaverse Post
Comments
0No comments yet.