# Google Cloud London outage: a power event in europe-west2-b and the UK services that wobbled

Published: 2026-10-20

A power event took out the europe-west2-b zone of Google Cloud's London region on Tuesday morning, London time. Google said services in that zone were unavailable and some services across the region saw elevated errors, while Revolut, HSBC, Lloyds Banking Group brands, DPD and the courts' remote-hearing video links reported problems. We walk through Google's own status updates, how availability zones are meant to contain this, Google's 2022 report on the same region, and two DeepSeek stories: a DHS order removing DeepSeek products from US federal civilian agencies and the open-source DeepInfer engine.

Canonical: https://thedailydiff.dev/video/2026-10-20-gcp-london-outage/

## What this video covers

- Google's incident gives a start of 07:08 London time on October 20, 2026; for hours it described only elevated errors and intermittent network failures in europe-west2, with the workaround listed as none for now.
- At 12:12 London time Google said a power event caused an outage within the europe-west2-b zone; services there were unavailable and some services elsewhere in the region saw errors. Thirty-eight products were listed on the incident.
- Google prioritised recovering existing workloads, warned new workloads might not be creatable yet, reported recovery for most workloads at 13:10 London time and called the issue mitigated for most customers at 15:03, promising an analysis. No root cause beyond the power event had been published when we checked.
- Revolut said a third-party issue caused its disruption and that money and data were safe; HSBC said the situation also affected other organisations. The BBC noted most firms had not officially linked their problems to Google.
- VMs and zonal disks live in one zone, so a single-zone deployment turns a zone failure into a full outage; spreading across zones trades it for lost capacity, at the cost of paying for spare capacity.
- In July 2022 a cooling failure in europe-west2-a also caused regional impact; Google's report said internal traffic was routed away from all three zones instead of one, leaving storage unable to reach some replicas.
- DHS ordered DeepSeek products, and third-party products developed using DeepSeek implementations, out of federal civilian agency IT; separately DeepSeek's Apache-2.0 DeepInfer engine supports DeepSeek-V4.1-Flash for offline inference only.

## Chapters

- 0:00 Can one London zone take down your bank app?
- 0:49 What did Google's status page actually say?
- 1:16 Why did one zone hit the whole region?
- 1:47 Who went down, and who said so?
- 2:15 Would your app have survived one dark zone?
- 2:49 Why is DeepSeek banned and trending on the same day?
- 3:44 What did London's last outage teach Google?
- 4:15 Verdict: does Google's London region need review?

## Transcript

### Can one London zone take down your bank app?

0:00 You think a cloud region survives losing one building. On Tuesday one Google zone in London lost power, and UK banking apps fell over. In this video: what did Google say? Why did one zone hit the whole region? And who had a backup? It's Tuesday, October twentieth, and this is The Daily Diff. And this same London region has failed like this before. Google wrote a report about it, and that report explains today.

0:23 I'll get to it at the end. Google Cloud is Google's business renting out servers, databases and storage. It's one of the three big clouds, next to Amazon and Microsoft. Its London region is called europe-west2, and it's split into three zones, so one zone can go dark while the other two keep running. That outage leads today's diff. Then the US government orders DeepSeek out of federal agencies,

0:45 and DeepSeek's own serving engine tops Hacker News. Here's Google's own status page.

### What did Google's status page actually say?

0:50 The trouble started at 7:08 in the morning, London time. For hours the only description was elevated errors and intermittent network failures in europe-west2. Under workaround, it said none for now. Every timestamp on that page is in Pacific time, for an outage in London. One Hacker News commenter noticed, politely. Later the workaround changed to this.

1:12 Move your workload to another zone or region, if feasible. At twelve minutes past noon, the cause.

### Why did one zone hit the whole region?

1:18 A power event took out the zone europe-west2-b. Google says its services were unavailable in that zone, and some services across the rest of the region saw errors too. Thirty-eight products are listed on the incident. First, Google said the power problem was mitigated. Then it put existing workloads ahead of new ones, and warned that customers might not be able to create new workloads yet. By ten past one most workloads were recovering, and just after three in the

1:43 afternoon Google called it mitigated for most customers. Then the customers.

### Who went down, and who said so?

1:48 Revolut told the BBC it had a service disruption caused by a third party, and that money and data were safe. HSBC apologised and said other organisations were affected too. Lloyds said its app was working, though some people had issues. More than fifteen hundred people reported Revolut problems to Downdetector shortly after eight. The courts service said its video links for remote hearings were hit. The BBC notes most firms haven't officially tied their problems to Google.

### Would your app have survived one dark zone?

2:15 Here's the part you control. A virtual machine or a zonal disk lives in exactly one zone. If your whole bank sits in zone b, the power event is your outage. Spread it across a and c, and you lose capacity instead of customers. That spare capacity costs money, which is why plenty of teams skip it. Monzo has written up the bigger version. Its main bank runs on Amazon's cloud, and a separate stand-in system on Google's can keep card payments and transfers going.

2:42 On Hacker News, one commenter guesses UK data rules pushed firms into one region. Nobody has confirmed that. Now the other story.

### Why is DeepSeek banned and trending on the same day?

2:49 Homeland Security has ordered DeepSeek out of federal civilian agencies, and the order lands in the Federal Register today. It covers DeepSeek's own products and also third-party products built using DeepSeek implementations. If you sell software to US agencies, read that clause before you ship a DeepSeek model inside your product. The day the ban hit the Federal Register, DeepSeek's newly open-sourced inference engine, DeepInfer, hit the top of Hacker News. DeepSeek says it's the engine behind its own production services,

3:18 under Apache 2.0. It runs one model, V4.1 Flash, on Nvidia's Blackwell data-centre GPUs or Huawei's Ascend chips, and only for offline inference. The KV cache system and the load balancing aren't in this release, so you get the engine without the rest of the car. The interesting switch is batch-invariant mode. Turn it on, and results stay identical no matter which other requests share the batch, on Nvidia only.

3:43 Now that report.

### What did London's last outage teach Google?

3:44 In July 2022, cooling failed in a data centre hosting zone a of this same region. Google admitted regional services were hit too. Its own traffic change steered internal services away from all three zones instead of one, so storage couldn't reach any copy of some data. Today it was power, and Google says a full analysis is coming. Back then, Google promised immediate steps to improve this region's resilience. Once again one zone went dark, and some services across the region threw errors

4:14 with it. My verdict is needs review.

### Verdict: does Google's London region need review?

4:16 A zone losing power is exactly what zones exist for. I want Google's report to explain why the rest of London wobbled again, and I want your bank to explain why one zone was enough. Tell me in the comments if you'd stamp it differently. If you'd rather read this than hear me say it, the diff lands in your inbox every morning, free at thedailydiff.dev, link below. And that's the diff for today. I'm Niko from Axrisi.

4:37 Merge responsibly.

## Sources

- [Multiple services are experiencing issues in the europe-west2 region (incident Xs2XTHzAVC9YnafnPyzj)](https://status.cloud.google.com/incidents/Xs2XTHzAVC9YnafnPyzj) — Google Cloud Service Health
- [Google Cloud says 'power event' behind outage hitting UK banks and websites](https://www.bbc.co.uk/news/articles/cxdd8mm6qy4mo) — BBC News
- [Google Cloud says outage affecting banks and court video links caused by 'power event' (live)](https://www.bbc.co.uk/news/live/c58jd2kjp93xt) — BBC News
- [Incident report: multiple Cloud products in europe-west2, July 2022](https://status.cloud.google.com/incidents/fmEL9i2fArADKawkZAa2) — Google Cloud Service Health
- [Regions and zones](https://cloud.google.com/compute/docs/regions-zones) — Google Cloud documentation
- [Tolerating full cloud outages with Monzo Stand-in](https://monzo.com/blog/tolerating-full-cloud-outages-with-monzo-stand-in) — Monzo
- [Google cloud outage takes down banks (discussion)](https://news.ycombinator.com/item?id=50163961) — Hacker News
- [Notice of Order Regarding Exclusion and Removal of DeepSeek Products (2026-21447)](https://public-inspection.federalregister.gov/2026-21447.pdf) — U.S. Department of Homeland Security · Federal Register
- [DeepInfer](https://github.com/deepseek-ai/DeepInfer) — DeepSeek on GitHub