Google Cloud Knocked Out UK Banks for 8 Hours Today

This morning at 06:08 UTC, a power failure in Google Cloud’s europe-west2-b zone in London knocked out Revolut, Lloyds, HSBC, Royal Mail, UK courts, and NHS-linked services simultaneously. The outage ran nearly eight hours, until 14:03 UTC. Google’s official explanation — issued hours into the incident — was four words: “a power event.” That vague shorthand is doing a lot of heavy lifting for what was, in practice, a single zone’s failure cascading into 38 cloud products and the daily lives of millions of UK residents.

Everything That Went Down This Morning

The blast radius was enormous for a single-zone failure. Revolut users could not make card payments, could not withdraw cash, could not log into the app. Lloyds and HSBC customers were locked out of funds. Halifax reported customers having “issues with our app and online.” Beyond banking, Royal Mail and DPD halted parcel operations, Rightmove went offline, BT and Screwfix reported disruptions, NHS England-linked services went dark, and UK court remote-hearing livestreams failed — including coverage of the Russell Brand trial proceedings.

Revolut confirmed “service disruption affecting multiple services” while noting that customer money and data remained safe. The Register’s coverage noted a Keele University lecturer’s assessment: “A problem in a single cloud region can therefore propagate through an entire digital supply chain.” Google’s status page eventually listed 38 affected products, including Cloud Run, GKE, App Engine, BigQuery Data Transfer, and Managed Apache Kafka.

The “Power Event” Explanation Is Not Enough

“Power event” is a term designed to sound technical while revealing nothing. It could mean a transformer failure, a cooling system malfunction, a voltage spike, or a backup generator that did not kick in. Google never said which. Compare this to their July 2022 europe-west2-a incident report, which explicitly named “a simultaneous failure of multiple, redundant cooling systems” and a routing misconfiguration that spread the failure regionally. That was the same London region — and it produced a more detailed post-incident account than what we have today.

Transparency here is not just good PR. Organizations running critical workloads in europe-west2 cannot make informed architecture decisions without knowing whether this was a one-off physical anomaly or a repeating design weakness in this region’s power infrastructure. Vague language protects Google’s reputation more than it helps its customers plan. Bloomberg’s reporting confirms the disruption extended well into the afternoon before full service was declared restored.

The Architecture Failure Behind the Google Cloud Outage

Zones exist precisely so a single data center’s bad morning does not take down your production service. Properly deployed multi-zone applications — spreading instances across europe-west2-a, -b, and -c with regional load balancing — should survive a zone-level failure. The evidence today suggests many of these organizations either deployed to a single zone, or had multi-zone configuration they had never tested under real failure conditions. The line circulating in developer communities discussing this outage is accurate: “A failover plan you have never run is a hope.”

Digital-first banks like Revolut are disproportionately exposed. Unlike Lloyds or HSBC, which still operate legacy on-premises systems as partial fallback, Revolut is cloud-native through and through. When europe-west2-b fell, Revolut went fully dark. Traditional banks also had outages, but had some layers between them and total service loss. Cloud-native architectures are faster to build, but the resilience risk is unambiguous when untested failover meets a zone-level power failure.

If you run workloads on GCP, run this audit now:

gcloud compute instances list

If everything shows europe-west2-b — or any single zone — you have the same exposure these organizations had this morning. Regional managed instance groups, regional load balancers, and cross-zone storage replication are not optional extras for anything carrying production traffic.

Key Takeaways

• Google Cloud’s europe-west2-b zone suffered a power failure at 06:08 UTC today, knocking out 38 products and cascading into outages for Revolut, Lloyds, HSBC, Royal Mail, NHS, and UK courts — lasting until 14:03 UTC.
• Google’s “power event” explanation is deliberately vague. This is the same region that suffered a major outage in July 2022 due to cooling failures and routing misconfiguration, and Google still has not explained today’s root cause clearly.
• Single-zone dependency is the real failure mode. Multi-zone deployment with tested failover would have kept most of these services running.
• Digital-first banks (Revolut) are more exposed than traditional banks — cloud-native architectures have no on-premises fallback when the cloud zone goes down.
• Audit your zone distribution now with gcloud compute instances list. An untested failover is not a resilience strategy.