Coinbase outage lasts more than five hours

Crypto exchanges sell a simple promise — the market is always there. That is the whole point. But late on May 7, Coinbase users suddenly could not trade, and the platform’s own status pages stayed in degraded mode for hours while the company worked through an Amazon Web Services problem. By the time Coinbase Exchange said all markets were re-enabled, more than five hours had passed. (status.coinbase.com) ### What actually broke? Coinbase first said at 5:56 p.m. PDT on May 7 that some customers might be unable to transact on web and mobile. At 6:53 p.m., it posted the same warning for Coinbase Exchange. By 7:37 p.m. on the main status page and 7:38 p.m. on Exchange, Coinbase had tied the trouble to an AWS outage and said customer funds were safe. (status.coinbase.com)? No — this went beyond a laggy interface. Coinbase Exchange moved through a controlled restart sequence, which tells you the matching engine itself had to be carefully brought back. First came notice that trading would be re-enabled shortly. Then all markets shifted to “Cancel Only” mode, where resting orders could be canceled but new market and l(status.coinbase.com)uction mode before full trading resumed. (status.exchange.coinbase.com) ### How long was Coinbase really down? If you use the first customer-impact timestamp, the outage started at 5:56 p.m. PDT on May 7. Coinbase Exchange said all markets were re-enabled at 12:49 a.m. PDT on May 8. That is 6 hours and 53 minutes from first warning to full market restoration. Even if some functions came back earlier, the trading venue itself was not fully normal again until after midnight on the West Coast. (status.exchange.coinbase.com) ### Why was AWS in the middle of this? The key detail is that Coinbase did not describe this as an internal software bug. Its status page said the disruption came from an AWS outage, and one later update pointed users to AWS service health for more information. Reuters described the broader event as an outage at an AWS data center in Northern Virginia that also aff(status.exchange.coinbase.com)base getting hit by a cloud dependency failure. (status.coinbase.com) ### Why does “Cancel Only” matter? Because it shows how exchanges try to avoid a messy reopen. Think of it like stopping traffic at an intersection, clearing the cars already stuck in the box, then turning the lights back on in stages. “Cancel Only” lets traders pull orders without adding fresh chaos. Auction mode then helps find an opening price before continuous matching re(status.coinbase.com)not in a normal market. (status.exchange.coinbase.com) ### Did funds go missing? Coinbase repeatedly told users their funds were safe. That matters, because exchange outages trigger two different fears at once — solvency and access. This incident looks like an access problem, not a custody shortfall. But for traders who needed to hedge, close positions, or move money in real time, “funds are safe” is not the same thing as “funds are usable right now.” (status.coinbase.com) ### Why does this land harder now? Because crypto is trying to look more like core market infrastructure. Coinbase is not a niche app anymore — it serves retail traders, institutions, APIs, and linked products across a market that now expects near-constant uptime. The awkward part is that even a giant exchange can still inherit very ordinary cloud fragility from a single provider zone or facility. (status.coinbase.com) ### Bottom line? The story is not just that Coinbase had a bad night. It is that a crypto exchange built for 24/7 markets was knocked out for nearly seven hours by infrastructure underneath it. That is the resilience question users keep coming back to — not whether funds were safe, but whether access is there when the market moves. (status.coinbase.com)