Coinbase restores trading after AWS outage

Crypto trading is supposed to feel continuous. Always on. Always there. But on May 7, Coinbase reminded everyone that a giant exchange can still go half-dark if the cloud under it stumbles. After hours of disruption tied to Amazon Web Services, Coinbase said all markets were re-enabled and customers could log back in and trade. ### What actually broke? Coinbase’s status page first flagged degraded performance at 5:56 p.m. Pacific on May 7. Later updates tied the disruption to “increased temperatures” in the affected AWS service, and Coinbase said the incident hit core trading systems badly enough that normal market activity had to be restricted. ### Why did trading stop? When an exchange loses confidence in the plumbing behind order handling, the safest move is to stop the parts that can create mismatches. (status.coinbase.com) Coinbase said it would begin re-enabling markets by first putting them into “Cancel Only” mode — basically, users could pull orders, but not fully resume normal trading yet. That is the exchange equivalent of reopening a highway one lane at a time after a pileup. ### How long was Coinbase down? The exchange’s incident history shows the degraded-performance event running from 5:56 p.m. PDT on May 7 to 2:28 a.m. PDT on May 8. That is about 8 hours and 32 minutes from first incident to resolution, even though some reports described the trading halt itself as roughly six to seven hours. The difference is useful — service incidents, market shutdowns, and full recovery are not always the same clock. (status.coinbase.com) ### Was this just one AWS server? Probably not. Coinbase-linked coverage and support updates pointed to failures affecting multiple AWS zones, not just one isolated machine. AWS’s public health page also showed an operational issue on May 8 involving EC2 and degraded EBS volumes in a single availability zone in us-east-1, with improvements only coming later. The clean takeaway is that even when the trigger is local — like overheating in Northern Virginia — the visible blast radius can spread through several dependent services. (status.coinbase.com) ### Why does that matter so much for crypto? Because crypto exchanges are not just websites. They are matching engines, risk systems, login flows, wallet operations, price feeds, and mobile APIs that all have to stay in sync. If one layer goes weird, the exchange cannot just shrug and let trading continue. A bad fill or a missing cancel is worse than a pause. That is why Coinbase kept repeating one reassurance during the outage — customer funds were safe — even while access and trading were not. (cointelegraph.com) ### Why did this land so hard this week? Timing. The outage hit right after a rough earnings stretch for Coinbase. Coverage around the incident noted a $394 million quarterly net loss, layoffs, and pressure on the stock, which made the platform failure feel less like a random bad day and more like another hit in a messy week. (status.coinbase.com) ### What does Coinbase do differently now? Coinbase has already been here before. After an AWS outage in October 2025, the company said it was reviewing regional deployment strategy and broader fixes to reduce dependence on the same failure patterns. So this week’s outage is not just an embarrassing blip — it is another proof point that resilience work is still unfinished. (msn.com) ### Bottom line Coinbase got trading back. That is the immediate story. But the more important one is structural — a market that sells itself as nonstop and decentralized still depends on very centralized infrastructure underneath. When that layer overheats, the whole promise gets stress-tested. (status.coinbase.com) (coinbase.com)