An AWS chiller unit failed, and Coinbase went down directly for 8 hours. Centralized exchanges’ Achilles’ heel is still too overly concentrated.

CryptoWorld News: Coinbase experienced a major outage lasting approximately 8 hours on May 7, 2026, with full recovery in about 12 hours. A cooling system failure in an AWS us-east-1 availability zone caused EC2/EBS to go offline, affecting multiple services. The matching engine lost a majority of nodes, losing quorum, requiring rebuilding the node group and code adjustments; a hosted Kafka control plane failure prevented automatic election of partition leaders, disrupting quotes and data streams. Recovery was achieved after manual partition migration. Coinbase stated it will enhance cross-region hot standby and disaster recovery drills.