Solana narrowly avoided a complete halt on Wednesday when a faulty BGP route (Border Gateway Protocol, the internet’s routing protocol) at Teraswitch in Miami disconnected nearly 29% of staked tokens. The episode highlights the fragility of the network infrastructure behind blockchains that market themselves as decentralized.
🔑 Key takeaways
- Roughly 29% of Solana’s staked tokens went offline simultaneously on Wednesday morning.
- The network sat ~20 million SOL from the total-freeze threshold, which triggers as soon as one third of tokens disappear.
- 90 validators hit, 333 SOL in lost rewards, fully covered by validator bonds.
- Origin: a bad BGP route at Teraswitch in Miami, propagated to London, Amsterdam, Frankfurt, Singapore and Tokyo.
- A single operator (AS2032) held more than 25% of staking, beyond Solana’s prescribed safety limit.
A safety mechanism that almost triggered
Solana embeds an automatic safeguard: transaction finalization halts as soon as more than one third of staked tokens drop offline simultaneously. On Wednesday, the network sat roughly 20 million SOL from that critical threshold, according to an analysis published by Marinade Finance, one of the leading liquid staking platforms on the network.
« If the delinquency had crossed one third, nothing finalizes for anyone holding SOL anywhere, and there is no guarantee that it would change. The February 2024 freeze took about five hours to restart. »
Marinade Finance, explanatory post
For context, the February 2024 incident immobilized the entire network for nearly five hours, cutting off traders and decentralized applications (DeFi) from any on-chain execution. This week’s episode proves the danger remains identical whenever a critical infrastructure provider wobbles.
Teraswitch: the source of the cascade
The initial point of failure lies in Teraswitch’s Miami data center. A faulty BGP route published by that operator cut connectivity for a large number of validators hosted in its facilities. The failure did not stop at U.S. borders: it propagated to data centers in London, Amsterdam, Frankfurt, Singapore and Tokyo. North America, broadly, remained operational throughout the incident.
| Data center location | Status during the incident | Impact duration |
|---|---|---|
| Miami (Teraswitch) | Origin of the outage | ~10 minutes to resolve |
| London | Disconnected | 33 minutes |
| Amsterdam | Disconnected | 33 minutes |
| Frankfurt | Disconnected | 33 minutes |
| Singapore | Disconnected | 33 minutes |
| Tokyo | Disconnected | 33 minutes |
| North America (rest) | Operational | None |
Teraswitch corrected the faulty route in ~10 minutes, and network traffic was fully restored at 4:16 UTC. Yet the affected validators remained offline for the full 33 minutes of the incident: their backup systems failed to activate automatically, prolonging the network’s exposure well past the operator-side fix.
Validator concentration: the real structural risk
Beyond the network outage, the episode exposes a blind spot in the Solana ecosystem: staking concentration among a small number of operators. A single actor, identified as AS2032, controlled more than 25% of all tokens locked to secure the network — above the safety limit prescribed by the blockchain’s designers. Almost all of those tokens dropped offline at once. Other operators lost 14 million additional tokens during the 33-minute window.
The majority of affected validators, including the large operator Helius, failed to switch over to their backup systems. The 90 affected validators collectively lost 333 SOL in rewards, an amount fully covered by their validator bonds, the collateral deposited by operators on Marinade.
$4.3 billion in DeFi at stake
At the time of the incident, Solana hosted roughly $4.3 billion in assets locked in decentralized finance protocols operating on its blockchain. For context, Solana has built its reputation as a faster, cheaper alternative to Ethereum, the sector leader, by combining Proof of Stake and Proof of History to process several thousand transactions per second.
That performance, however, comes at a cost: the network has experienced several notable outages in recent years, including the February 2024 incident mentioned above. Wednesday’s episode confirms that the network’s resilience depends as much — if not more — on the diversity of its infrastructure providers as on the strength of its protocol.
Conclusion: decentralization remains a work in progress
Wednesday’s alert lasted only 33 minutes, but it exposed two structural fragilities in Solana. First, the dependence on a small number of connectivity providers such as Teraswitch, where a single routing error can destabilize a significant fraction of the network. Second, the excessive concentration of validators among a limited number of operators, beyond the blockchain’s own internal safety thresholds.
In the short term, attention will turn to diversifying network routes and systematically activating failovers (automatic backup systems) at professional validators. In the medium term, the Solana Foundation and operators like Marinade will need to convince the community to tighten anti-concentration rules — otherwise the risk of a total freeze, built into the architecture itself, will continue to loom over every activity spike.
Sources
- CoinDesk — Solana nearly froze Wednesday
- Marinade Finance — Terms of use
- CryptoNews — DeFi section
- Yahoo Finance — Solana DeFi incidents
- TradingView — Step Finance shutdown
This article is published for informational and educational purposes only. It does not constitute investment advice. Do your own research (DYOR) before making any decision.
