Google Cloud is having a bad Tuesday in Europe. Since around 07:00 BST this morning, multiple Cloud services in the europe-west2 region have been suffering network failures, according to Google's own Cloud Status page — which says its engineering team is still investigating. The biggest name affected so far is Revolut, whose customers started the day unable to use the bank's services normally.
If you run anything on Google Cloud in Europe, here is what is confirmed so far, and the track record it adds to.
What is happening right now
Google's status message is brief: "We are experiencing an issue with multiple services in europe-west2," beginning around 07:00 BST (23:08 PDT the previous evening). No official cause has been announced yet — this is still an active incident.
Image: screenshot of Google's public Cloud Service Health dashboard, taken 20 October 2026; incident details via the Google Cloud status dashboard and BBC live coverage.
Revolut reported a "service disruption affecting multiple services" this morning, confirming that customers' money and data are safe and saying it has been working on the fix since the issue was identified. Delivery company DPD is also showing up among services with reported problems, though the BBC notes it is not confirmed that every company on the DownDetector list is a Google Cloud customer. Correlation is not causation — but when a cloud region and a stack of European services stumble at the same hour, it is usually the same hour.
One jurisdictional footnote: europe-west2 is Google's London region — the UK, post-Brexit, rather than the EU. The europe-naming confuses people, but the Paris (europe-west9), Frankfurt (europe-west3) and Belgian (europe-west1) regions are the EU-home ones. Regardless, europe-west2 serves plenty of EU customers, and Europe feels it.
GCP's European outage history
This is not a first. From Google's own incident history, the European entries stand out:
- July 2026, Netherlands (europe-west4) — the worst recent one. A voltage transient from the utility grid tripped breakers, backup power partially failed, and the chiller controls went offline. Data halls hit 44°C, forcing Google to shut down equipment to protect it. VMware Engine, Bare Metal Solution and NetApp Volumes were down for the better part of 15 hours. Twenty GCVE private clouds and nine bare-metal customers were directly affected.
- June 2025, multi-product — around 3 hours of service issues across Firestore, Data Fusion, Looker Studio and monitoring products.
- May 2025, Compute Engine — a GCE issue cascading into dependent services across zones for nearly 9 hours.
- February 2024, Milan (europe-west8) — multi-product outage lasting about 2 hours.
- April 2023, Paris (europe-west9) — networking and compute problems in the zone stretching over a day.
- June 2023, global — operational latency across many GCP services for over 4 hours.
The July Netherlands incident deserves attention beyond its duration: it was a physical cascade — power, then cooling, then heat, then deliberate shutdowns. No amount of software redundancy fixes a 44-degree data hall. And it hit precisely the services (VMware migration, bare metal databases) that risk-averse enterprises buy to feel safe.
What EU businesses should take away
First, the boring and correct lesson: every cloud provider fails. AWS, Azure and Google all have multi-hour incidents every year. The fix is not a different logo, it is architecture — multi-region deployments, failover, and backups you have actually tested, which Google itself recommends in every incident report.
Second, geography is not redundancy. Running in "Europe" is one region failing at a time; europe-west2 going down while europe-west3 stays up is the design working as intended. The customers who went dark are the ones who built in one place.
Third, this is also why the EU's sovereignty conversation keeps returning. When a single London-region incident ripples into banks and logistics across the continent the same morning, "which jurisdiction runs our infrastructure" stops being an abstract policy debate — see our map of European cloud and hosting alternatives. Europeans do fail too (their incident pages exist), so the honest pitch is diversification and jurisdiction you can read, not miracles.
We will update this post when Google publishes its incident report with a root cause.
No comments yet