On the evening of May 7, 2026, Coinbase went dark. Users across web and mobile platforms found themselves locked out of their accounts – unable to trade, transfer funds, or even view their balances. The outage lasted approximately five to six hours. The cause: a cascading failure at an Amazon Web Services data centre in Northern Virginia.
The incident wasn’t just embarrassing for one of crypto’s most prominent exchanges. It exposed a structural vulnerability that critics have pointed to for years – the irony of a decentralised financial system running on the most centralised infrastructure stack imaginable.
What Happened
At approximately 5:25 PM PDT on May 7, AWS engineers detected problems in a single availability zone – use1-az4 – within AWS’s US-EAST-1 region, its oldest and largest US hub in Northern Virginia. An overheating incident at the facility cascaded across multiple availability zones, degrading or severing connections for services that relied on them.
Coinbase acknowledged degraded performance at around 6:53 PM PDT. Roughly 45 minutes later, the company publicly attributed the issue to AWS’s infrastructure failure. By midnight, service had largely been restored, though some users reported lingering issues into the following morning.
Coinbase CEO Brian Armstrong characterised the outage as “never acceptable” in a public statement, while noting that the failure hadn’t affected most of the company’s internal design systems – suggesting some redundancy had held up, even as customer-facing services collapsed.
The Centralisation Paradox
The incident arrived at a particularly uncomfortable moment. Coinbase had just reported its Q1 2026 earnings – a quarter in which the company posted a net loss while revenue came in below analyst expectations. The simultaneous outage drew sharp criticism from commentators who questioned the company’s infrastructure priorities.
Gergely Orosz, an engineer and author of The Pragmatic Engineer newsletter, put it bluntly on social media: “This outage is because Coinbase seems to have a hard dependency on AWS, and when AWS (or a part of it) is down, so is Coinbase.”
The observation lands harder when you consider what Coinbase is. it’s one of the world’s largest crypto exchanges – an institution that markets itself to retail and institutional customers alike on the premise that it provides reliable access to a decentralised financial system. That system now apparently runs on a single cloud provider’s aging data centre infrastructure in Virginia.
Not an Isolated Problem
Coinbase isn’t alone. A significant portion of the crypto industry’s infrastructure – exchanges, custodians, data providers, oracle networks, and even some validator sets – runs on AWS, Google Cloud, or Microsoft Azure.
This dependency has been flagged repeatedly as a systemic risk. The Ethereum Foundation, for example, has published data showing that a disproportionate share of validator nodes run on AWS. A major cloud outage, if it hit during a period of high network activity, could theoretically affect block production, MEV, and transaction finality in ways that most users don’t expect.
The Coinbase outage is a reminder that “decentralised” as a technical description of a blockchain’s consensus mechanism says nothing about the centralised infrastructure that most people use to interact with it.
What Coinbase’s Architecture Tells Us
The fact that a single AWS regional failure could take down Coinbase’s customer-facing platforms entirely suggests the exchange lacks sufficient multi-region or multi-cloud failover for its critical trading and account management services.
Best-practice resilience engineering for financial infrastructure typically requires active-active or active-passive deployments across at least two cloud providers or two geographic regions, with automatic failover that limits customer impact to seconds, not hours.
The five-hour duration of the Coinbase outage suggests its failover configuration either didn’t exist at the required scale or didn’t function as designed.
Staff Layoffs Add to the Picture
The timing of the outage drew additional scrutiny because Coinbase had recently announced another round of staff reductions. Critics on social media quickly connected the dots, suggesting that infrastructure engineering headcount cuts may have reduced the team’s ability to maintain or respond to complex multi-region failover scenarios.
Coinbase hasn’t publicly addressed whether the layoffs affected its infrastructure team or whether staffing played any role in the outage response time.
What the Industry Should Take From This
The AWS outage affecting Coinbase won’t be the last incident of its kind. As the crypto market grows more institutional and more retail investors move significant portions of their savings onto digital asset platforms, the stakes of infrastructure failures rise.
Several concrete steps have been proposed by infrastructure specialists in the wake of the event:
- Multi-cloud redundancy: Critical trading infrastructure should operate across at least two cloud providers with automatic failover tested quarterly
- Regional distribution: Deploying across multiple geographic cloud regions, not just multiple availability zones within a single region
- Chaos engineering: Regular adversarial testing that simulates single-provider failures
- On-premise fallback capability: Maintaining the ability to run core operations without any cloud dependency for short periods
Whether Coinbase and its peers will invest the resources required to close these gaps remains to be seen. The company’s Q1 loss suggests it isn’t operating from a position of financial comfort. Infrastructure upgrades at the required scale are expensive.
FAQ
Why was Coinbase down for so long if other companies survived the AWS outage? Coinbase appears to have had a hard dependency on a single AWS region without sufficient automated failover. Companies with better multi-cloud or multi-region architecture saw their services recover much faster or not fail at all. The five-hour duration suggests the recovery required significant manual intervention.
Can decentralised exchanges (DEXs) avoid this problem? Partially. Decentralised exchanges like Uniswap or dYdX run their core logic on blockchains, which aren’t dependent on any single cloud provider. However, their front-end interfaces – the websites users actually visit – typically run on centralised cloud infrastructure. An AWS outage could take down the user interface even if the underlying smart contracts remain accessible.
Is my money at risk when Coinbase goes down? An outage of this kind doesn’t put funds at risk of loss – funds remain on the blockchain and in Coinbase’s custody systems. However, the inability to trade during a volatile market period can result in real financial harm if users can’t execute stop-losses or respond to price movements.
*Sources: CoinDesk, CCN, CrowdFund Insider, CryptoTimes, Stocktwits, PaymentExpert. Incident date: May 7, 2026.*
