AWS outage knocks Coinbase, CME

Cloud outages are usually annoying. This one hit the plumbing of markets. AWS said a Northern Virginia data center overheated on May 7, knocking out part of one availability zone in us-east-1 and disrupting customers that happened to sit in the blast radius. Coinbase tied its own outage directly to that event, while CME reported technical and latency issues during the same window. (money.usnews.com) ### What actually broke at AWS? The immediate problem was heat. AWS said a rapid temperature spike at a single Northern Virginia data center led to a power loss, which then impaired infrastructure in one availability zone. That matters because “one zone” sounds na(money.usnews.com)WS shifted traffic away from the affected zone for many services, but also warned that full recovery for the remaining impaired systems would take hours. (tech.yahoo.com) ### Why did Coinbase go down so hard? Coinbase’s own status page made the link pretty explicit. It said customers were seeing degraded performance because of an AWS outage, then later said service disruptions were due to increased temperatures in the affected AWS service. Before reopening, Coinbase said markets would(tech.yahoo.com)diately throwing the whole venue back into normal matching. By 02:28 PDT on May 8, Coinbase said it would begin re-enabling trading, and later marked the incident resolved. (status.coinbase.com) ### Was CME hit the same way? That part is murkier. Reuters said CME Group faced technical and latency issues on its trading platform during the AWS incident, but CME did not publicly pin the cause on AWS the way Coinbase did. Its Global Command Center page later showed no current issues, which tells you the disruption had cleared, not necessarily what caused it. So(status.coinbase.com)E was disrupted in the same window, and the exact causal chain for CME was less clearly spelled out in public. (money.usnews.com) ### Why does one availability zone matter so much? Because “multi-cloud” and “high availability” often sound stronger than they are. A lot of systems are redundant across servers but not across true failure domains. If matching engines, APIs, databases, or market (money.usnews.com)ustering, but the catch is that tight clustering can also mean shared physical risk. This outage is a clean example of that tradeoff. (itpro.com) ### Why is this extra awkward for Coinbase? Timing. The outage landed the same day coverage around Coinbase was already focused on weak quarterly results and internal cuts, so the downtime fed a broader argument that the company is under pressure operationally as well as financially. That does not mean the outage was Coin(itpro.com)ture when they cannot trade. (coindesk.com) ### Is this just a Coinbase story? No — it is a market-structure story. When a cloud zone failure can interrupt crypto trading and rattle a major derivatives venue on the same day, the lesson is bigger than one exchange. Infrastructure placement is not just about speed anymore. It is about making sure the fastest setup is not also the most fragile one. (money.usnews.com) ### Bottom line? A cooling problem in one AWS zone was enough to disrupt real trading activity. That is the part firms will remember. Redundancy that looks solid on a diagram can still fail if too much of the business lives inside the same cloud blast radius. (tech.yahoo.com)