Google Cloud outage in London, October 20, 2026

20th Oct 2026

Multiple Google Cloud services are down or degraded in Google's London region, europe-west2, after a power event in the europe-west2-b zone. The incident began at 11:08 PM Pacific on October 19, 2026 (2:08 AM Eastern and 7:08 AM London time on October 20), according to Google's service health dashboard. Google says Cloud services are unavailable in the europe-west2-b zone, and some services across the europe-west2 region may see elevated errors and intermittent network issues. The dashboard lists 38 services in scope, including Compute Engine, Kubernetes Engine, Cloud Storage, BigQuery, Cloud SQL, Cloud Pub/Sub, Cloud Run, Cloud Load Balancing, and Cloud Firestore.

Google's 4:12 AM Pacific update (7:12 AM Eastern, 12:12 PM London) said the power event itself is mitigated, partial mitigation for infrastructure services would be observed within 30 minutes, and the remaining services would be mitigated after that. Google advised customers who can do so to migrate affected workloads to another zone or region.

UPDATE, October 20, 2026, 4:58 AM Pacific (7:58 AM Eastern, 12:58 PM London): Google says it is seeing the recovery of services and is prioritizing the recovery of existing workloads. Customers may not yet be able to create new workloads. Google says full recovery of existing workloads should be completed tentatively by Tuesday, October 20, 2026, 06:00 (all incident page times are US/Pacific), and that it will provide a further update by 6:00 AM Pacific. No compensation has been stated. The incident remains Active; it is not yet restored. Downstream customer Collibra, reporting through its incident record, said at 09:33 UTC that its environments hosted in GCP europe-west2 were in recovery with over half already up and accessible, with remediation still in progress.

UPDATE, October 20, 2026, 5:44 AM Pacific (8:44 AM Eastern, 1:44 PM London): Google says it is seeing the recovery of services for most workloads in europe-west2 as of 5:10 AM Pacific (8:10 AM Eastern, 1:10 PM London). Google will provide a further update by 7:00 AM Pacific (10:00 AM Eastern). The symptoms line is unchanged: services remain unavailable in the europe-west2-b zone, and some services in the region may still see elevated errors and intermittent network issues. Google has not marked the incident resolved, and no compensation has been stated. Reuters reported at 12:43 UTC that Google Cloud services were largely back up in the UK, consistent with Google's most-workloads wording; the incident remains Active on Google's own record, which is the primary source for restoration.

UPDATE, October 20, 2026, 7:03 AM Pacific (10:03 AM Eastern, 3:03 PM London): Google closed the incident. In its final update Google said the issue with the impacted products has been mitigated for most customers and is believed to be affecting currently a very small number of customers, and that its engineering team is working on that remainder. Google said no further updates will be provided on the incident record, that customers who are still impacted should open a support case, and that it will publish an analysis of the incident once its internal investigation is complete. The workaround line now reads: none, as the issue has been mitigated. Google's dashboard lists the incident as Closed at 7:05 AM Pacific (10:05 AM Eastern, 3:05 PM London). No compensation has been stated. Reuters reported that Google said full service had been restored in the London region; that phrasing is broader than Google's own final wording, which is the wording this page follows: mitigated for most customers, with a very small number still affected at closure.

Timeline:

October 19, 2026, 11:08 PM Pacific / October 20, 2:08 AM Eastern / 7:08 AM London: incident began; power event in europe-west2-b zone.
October 20, 2026, 4:12 AM Pacific / 7:12 AM Eastern / 12:12 PM London: Google said the power event is mitigated and partial mitigation was in progress; no restoration time published at that point.
October 20, 2026, 4:58 AM Pacific / 7:58 AM Eastern / 12:58 PM London: Google reported recovery of services underway, existing workloads prioritized, new workloads not yet creatable, tentative full recovery of existing workloads by 06:00 Pacific, next update by 6:00 AM Pacific.
October 20, 2026, 5:44 AM Pacific / 8:44 AM Eastern / 1:44 PM London: Google reported recovery of services for most workloads in europe-west2 as of 5:10 AM Pacific; next update by 7:00 AM Pacific. Incident still listed as a service disruption, not resolved.
October 20, 2026, 7:03 AM Pacific / 10:03 AM Eastern / 3:03 PM London: Google's final update: issue mitigated for most customers, believed to affect currently a very small number of customers, engineering team working on the remainder, no further updates on the record, post-incident analysis to be published. Dashboard closed stamp 7:05 AM Pacific / 10:05 AM Eastern / 3:05 PM London. Start to closure: 7 hours 57 minutes.

Downstream impact: Revolut, a UK digital bank and Google Cloud customer, reported a simultaneous service disruption on October 20. Reuters reported its status page showed most services unavailable or partially unavailable, including card payments and ATM withdrawals, and a person familiar with the matter linked Revolut's outage to the Google Cloud disruption; Reuters' later copy noted the link was not yet confirmed and that Revolut uses Google Cloud services. Downdetector spikes for other UK banking and delivery apps were reported in UK coverage, but Lloyds said its app was not down, so those reports are not counted here as confirmed impact.

Status: Closed by Google on October 20, 2026, at 7:05 AM Pacific (10:05 AM Eastern, 3:05 PM London), after 7 hours 57 minutes; mitigated for most customers, with a very small number still affected at closure on Google's own wording. At last check, October 20, 2026, 10:53 AM Eastern. This is a power event, not a cyber incident; no indication of a cyberattack has been reported.

Sources: Google Cloud Service Health, Reuters, Google Cloud services largely back up in UK, Reuters, AppleInsider, Collibra incident record via Pingoru