A thermal failure inside an Amazon Web Services facility in Northern Virginia knocked out more than eight cloud services on May 7, disrupting crypto trading, derivatives markets, and project management tools worldwide. Coinbase, one of the largest US-regulated crypto exchanges, bore the sharpest impact: core trading was suspended for over five hours during an already damaging day financially.

What Happened

Rising temperatures inside a single data center in the use1-az4 availability zone of AWS's US-EAST-1 region triggered power disruptions to hardware racks beginning May 7. AWS deployed additional cooling capacity by 10:11 PM PDT that evening, but recovery stretched into May 8 and moved slower than the company had projected. AWS's own status page began timing out during the incident, leaving developers unable to get reliable information about which services were still running. One affected customer described the situation to The Register: "All my servers in that region have just gone inaccessible, and AWS Dashboard is misbehaving, with even the status page timing out."

Among the primary confirmed services affected were EC2 compute instances, EBS storage volumes, ElastiCache, Managed Streaming for Apache Kafka, OpenSearch, and SageMaker. Elevated error rates cascaded across dependent services, bringing the total count of impaired services above eight. CME Group derivatives trading, FanDuel (the sports-betting platform), and Atlassian's Trello project tool were also disrupted.

Coinbase's Rough Day

Coinbase briefly entered "cancel only" mode during the disruption, meaning users could cancel existing orders but not place new ones, before the platform transitioned through broader degraded performance states across the five-hour outage. Solana and ALEO token transfers were specifically delayed. Fiat services kept running and customer funds were not at risk. The exchange said in a statement: "Coinbase systems are designed to be resilient to a single zone outage. In this case, we observed failures impacting multiple AWS zones, which caused an extended outage of core trading services."

The timing was damaging beyond the technical failure itself. On the same day the outage extended into May 8, Coinbase reported Q1 2026 earnings showing revenue of $1.41 billion against analyst expectations of $1.52 billion, a 31% year-over-year decline. The company posted a net loss of $394 million, or $1.49 per share, including $482 million in unrealized losses on held crypto assets. Shares dropped more than 5% in after-hours trading. Three days earlier, on May 5, Coinbase had announced a 14% workforce reduction affecting roughly 660 employees. Tech commentator Gergely Orosz, writing to an audience of more than 310,000 followers, noted the "unfortunate timing for Coinbase to have an hours-long outage when customers couldn't trade, days after their CEO said how non-technical teams are shipping code to production."

This was also the second major AWS-related Coinbase outage within months. A prior disruption hit the exchange in October 2025.

A Pattern, Not an Anomaly

US-EAST-1 is AWS's oldest and largest region, and also its most incident-prone. A DNS and networking failure in December 2021 took large sections of the internet offline. An October 2025 outage lasting 15 hours was caused by a race condition in DynamoDB's automated DNS management, affecting more than 70 services. AWS has attributed the region's frequency of incidents to its greater operational scale rather than any structural weakness, a position that critics consider inadequate.

This thermal event is also the second cooling-related AWS failure in recent months. In February 2026, a fire originating in the cooling system of AWS's Abu Dhabi data center (me-south-1), triggered by outdoor temperatures exceeding 38 degrees Celsius, caused a 36-hour outage affecting cloud services across the Middle East, South Asia, and Africa.

The May 2026 Virginia outage carried its own direct regional consequences outside the United States as well. Several major Indian crypto platforms, including CoinDCX, WazirX, and CoinSwitch, route their infrastructure through US-EAST-1 rather than the geographically closer AWS Mumbai region (ap-south-1), leaving them exposed to the full duration of the disruption. Indian developers building on Solana experienced RPC failures as servers routed through Virginia went offline. In Africa, traders in Nigeria, Kenya, and South Africa encountered residual service disruptions during early-morning local hours on May 8. Kenya's BitKE had previously flagged this kind of regional cloud dependency as a structural risk for African crypto infrastructure, and the Virginia outage gave that warning concrete form.

The Centralization Problem, By the Numbers

Blockchain networks continued settling transactions throughout the Virginia outage. Neither Bitcoin nor Ethereum halted. The disruption lived entirely at the application layer: exchange frontends, wallets, RPC endpoints (the servers that relay user requests to blockchain nodes), and custody APIs. That distinction matters, but it does not resolve the deeper problem.

According to Ethernodes data, approximately 2,368 Ethereum execution-layer nodes, around 36% of the total, run on AWS infrastructure. Messari research estimates that 55% to 80% of validator and node infrastructure across major proof-of-stake networks is concentrated among just four hosts: AWS, Hetzner, OVH, and Google Cloud. Two MEV builders, the software that orders transactions inside Ethereum blocks, account for more than 85% of all blocks produced on the network. Rotki founder Lefteris Karapetsas summarized the contradiction bluntly: "The whole vision behind blockchain was decentralised infrastructure, which we have completely failed on."

Filecoin and Arweave, two decentralized storage tokens, showed relative price strength in the 24 hours following the outage.

What Comes Next

AWS explicitly advised affected customers to migrate workloads across multiple availability zones and restore from EBS snapshots, a reminder that single-zone deployments offer no redundancy. For developers in South Asia and Africa building on Solana RPC providers like Helius and QuickNode, or relying on MetaMask via Infura, all of which route through AWS-hosted infrastructure, the guidance is the same: audit your zone distribution now. A decentralized RPC alternative from Consensys and EigenLayer, launched after the October 2025 outage, offers one concrete option worth evaluating. The February Abu Dhabi fire and the May Virginia thermal event, arriving within three months of each other, point to what researchers characterize as a broader, systemic challenge with thermal management across the cloud infrastructure that crypto builders depend on.