Google Cloud’s London Power Cut Tests New Bank Oversight

A power cut in Google Cloud’s London zone europe-west2-b knocked 38 products offline for 7 hours 52 minutes on 20 October 2026. Revolut, several high-street banks, parcel sites and remote court video all ran into trouble in the same window. Google restored the region that evening and blamed a power event at a third-party facility.

The UK had already put Google Cloud under direct finance oversight on 13 July, after the Treasury warned that an outage at one of a handful of suppliers could hit many firms at once.

The Outage Ran 7 Hours 52 Minutes From Zone B

Google’s public clock starts at 07:08 UK time, logged as 23:08 US/Pacific on 19 October. The incident log for the London region marks the event closed at 07:00 US/Pacific on 20 October, which is 15:00 in Britain. That span is 7 hours 52 minutes.

Early notes said multiple services in europe-west2 were seeing more errors than usual and network faults that came and went. There was no workaround. Engineers later pinned the fault on a power event inside europe-west2-b, one of three availability zones in the London region. Google said the power problem itself had been contained, then spent the early afternoon bringing existing jobs back before it would let customers start new ones.

THE MORNING ON THE STATUS PAGE

1. 20 October, 07:08 UK: Fault begins in europe-west2. Google later timestamps this as 23:08 US/Pacific on 19 October.
2. About 08:00 UK: Revolut customers start hitting dead cards, failed logins and broken cash machines, roughly an hour after the cloud region failed.
3. 12:12 UK: Google says a power event caused an outage in europe-west2-b and that the power event is contained. It tells customers they can try another zone or region if they can.
4. 13:10 UK: Most workloads in the region are coming back. New jobs may still fail.
5. 15:00 UK: The dashboard closes the incident. A very small number of customers are still affected, and Google says it will publish an analysis after an internal investigation.
6. Tuesday evening: A Google Cloud spokesperson says the company has resolved a disruption at a third-party facility in europe-west2 due to a power event, and that services have been restored.

The unnamed hall is the detail that sits under the brand name. This was not a cyberattack and not, on Google’s telling, a failure in a building it runs itself. A landlord’s power feed in one zone was enough to take Compute Engine, Kubernetes Engine, Cloud SQL, Cloud Storage, BigQuery and 33 other listed products with it. Looker Studio, Google noted, was not hit.

Google’s Zone Guide Exists for Power Cuts Like This

London is not a single room. Google publishes europe-west2-a, europe-west2-b and europe-west2-c as separate deployment areas in the same region, with fat pipes between them. The point of that split is written in plain language on the public putting resources in different zones guide: it reduces the risk of an infrastructure outage hitting every resource at once.

The same page is blunter about electricity. Google designs zones to cut the chance of shared failures from power, cooling or networking, so that if one zone drops you can shift traffic to another zone in the same region and keep running. Spread across regions, it says, and you get a still higher degree of independence when a whole region fails.

That is the architecture the status page then asked customers to use in the middle of a live incident. For anyone whose production estate already sat in more than one zone, the morning was a failover drill. For anyone whose customer-facing rails lived in europe-west2-b, the drill was the outage.

Banks, Courts and Parcel Sites Failed in the Same Window

Revolut was the firm that said the quiet part out loud. A person familiar with its stack had already tied the app failure to the Google Cloud disruption, and Revolut later confirmed the link. Other names showed up in user reports and in their own holding statements without always naming the cloud underneath.

WHO LOST SERVICE ON 20 OCTOBER

An outage tracker logged more than 1,500 Revolut reports shortly after 08:00 UK time, and also showed spikes for Lloyds, HSBC, Rightmove, DPD, NHS England and Royal Mail. Forest, the e-bike operator, was among the consumer apps caught in the same window. A house purchase stalled when a mortgage sat in a broken queue. A pub manager went back to pen and paper.

Lloyds said later on Tuesday that it understood the root issues had been resolved and that it expected all services back to normal soon. HSBC said it was aware some customers were having trouble getting in. Those are narrower claims than a blanket “the banks went down,” and they matter, because a high-street app that stays up while a digital bank’s cards die is a clue about who actually sat in zone b.

ING Germany had a separate login fault the same morning. There is no public link between that incident and Google.

Revolut Doubled Down on Google Cloud in 2025

Revolut is a London fintech. It already ran core systems on Google Compute Engine, and it is one of Google Pay and Wallet’s largest partners. On 4 September 2025 the two firms announced a multi-year and multi-million dollar partnership to carry that stack through a push past 100 million customers. Revolut then had more than 60 million customers, up from 1.5 million in 2018, and said those people were doing more than half a billion transactions a month. It now serves more than 80 million personal users.

The 2025 deal was not only more servers. Revolut said it would use Google’s AI tools, including Gemini models, to improve fraud detection and to shape product offers. David Tirado, vice president of global business and profitability, put the bet in one line.

Google’s world-leading AI and cloud infrastructure are an incredible asset. This partnership will empower us to rapidly scale our offerings, deliver even more personalised and valuable experiences to our customers, and continue to outpace the traditional financial sector. We’re building the future of finance, and Google is a key partner on that journey.

David Tirado, Vice President of Global Business and Profitability, Revolut, 4 September 2025

Tara Brady, president of Google Cloud EMEA, called the same kit secure, scalable and intelligent. On Tuesday morning that kit could not take a card payment, spit out cash, move a transfer or open the app. Revolut told customers their money and data stayed safe, which is the line a regulated e-money firm has to be able to say. Safety of balances is not the same as being able to spend them at 08:15 on a Tuesday.

Users in the UK, Ireland, France, Spain, Romania and Germany reported the break. Calling that a brief glitch does not survive contact with a full working day of declined cards. People who needed to pay were left refreshing a status page and a social feed, and the company’s own restore note did not land until 17:39 UTC.

Due to a recent Google Cloud outage, customers may have experienced temporary service disruption on the Revolut app today.

The issue has now been fixed, and services are now restored.

All our customers’ money and data is safe, and has remained safe throughout this incident. We…

— Revolut (@Revolut) October 20, 2026

The post is the record Revolut chose to put on the record: Google Cloud outage, services restored, money and data safe, apology. It does not say why a bank of this size still had customer-facing payments on a path that died with one London zone.

What the UK’s Critical Third Party Powers Cover

HM Treasury named Amazon Web Services EMEA SARL, Google Cloud EMEA Limited, Microsoft Ireland Operations Ltd and Oracle Corporation UK Limited as the first firms under a new overlay, and the Bank of England, the Prudential Regulation Authority and the FCA began to first Critical Third Parties on 13 July. The Treasury’s concentration finding was that more than 65% of UK organisations rely on those four for infrastructure, and that a failure at any one of them could hit many firms at once.

The label is a supervisory status, not a penalty. It sits in the Financial Services and Markets Act 2000 and lets the three watchdogs write rules for the systemic services these companies sell into finance, pull information, and take enforcement action. None of the four is a bank. Until July, a regulator could lean on a licensed firm and still have no direct hold on the supplier that firm rents its computers from.

WHAT THE OVERSIGHT CAN DEMAND

• Resilience tests: The Bank, PRA and FCA can put each designated provider through testing of the services it sells into UK finance.
• Self-assessments: They can require regular written assessments of how those services would hold up if a site, zone or region failed.
• Incident reports: They can order the firms to report major incidents rather than leave banks to pass the news up after customers start shouting.
• Enforcement: They can take action against a designated provider over the systemic services it provides to regulated firms and financial market infrastructure.

Sarah Breeden, the Bank’s deputy governor for financial stability, said the point on the day the names went up.

As critical third parties become increasingly embedded in the operations of financial institutions, they can introduce new forms of systemic risk. Our proportionate approach to overseeing these providers will ensure that these dependencies are managed in a way that safeguards financial stability.

Sarah Breeden, Deputy Governor for Financial Stability, Bank of England, 10 July 2026

EU supervisors had already named 19 technology firms, Google Cloud among them, as critical providers to European finance under the Digital Operational Resilience Act in November 2025. Oracle did not make that EU list. The UK added it. Tuesday was 99 days after the UK regime went live, and it was the first morning the new file had a live incident in it.

If Feasible Was the Only Workaround Posted

Once Google had a cause, the status page offered one path out: if feasible, move the affected workload to another zone or region. That sentence is a shrug dressed as a runbook. A payments stack cannot be dragged to Frankfurt between the morning commute and the first declined coffee. Data-residency rules, latency, and the simple fact of what is already in production all sit on the wrong side of “if feasible.”

Google’s own guide had already said the quiet part. If you host everything in one zone and that zone goes offline, you will not be able to reach any of it. Zone b went offline. Revolut’s cards, cash machines, transfers, FX and trading went with it. The high-street banks that only lost slices of online service look, from the outside, as if more of their estate sat somewhere else, or as if the Google region was not on the path for every feature. That difference is the whole design argument, and it is now a fact pattern in a regulatory file rather than a slide in a cloud-architecture talk.

The other uncomfortable layer is the building. Google called it a third-party facility. The Bank of England’s CTP note warns that disruption at a designated provider can become a single-point-of-failure impacting multiple firms at the same time. Tuesday added a rung: the designated provider was itself sitting in someone else’s hall, on someone else’s power. The name of that operator has not been released. The post-incident analysis Google promised is where that chain has to be written down, including why backups in that zone did not keep europe-west2-b up.

Hillier Gave Bailey and Rathi Until 4 November

An FCA spokesperson said the regulator was aware of an incident affecting services provided by Google and was engaging with Google and with affected firms to understand the impact. People need to be able to access their money and make payments, the spokesperson said, so the FCA expects firms to fix issues quickly, keep customers informed, and make sure no one is left out of pocket. The Bank of England and the PRA were also in contact with Google under the new overlay.

Dame Meg Hillier, chair of the Treasury Select Committee, wrote on Tuesday to Andrew Bailey, the Bank’s governor, and Nikhil Rathi, the FCA’s chief executive, asking how they responded and what they will do next. She asked for an initial reply by 4 November and a full reply three months after that, and she raised compensation for customers who could not get to their money.

Google said it would publish an analysis of the incident once its internal investigation is done. Revolut’s app was back the same evening, with money and data still where they should be. The open question is no longer whether europe-west2-b can go dark. It did. The file now sitting with Bailey and Rathi is whether a payments firm with more than 80 million personal users, on a deepened Google Cloud deal, can still lose cards and cash machines when one London zone loses power, 99 days into a regime built to stop that becoming a system event.