Why a Google Cloud outage can take down banking apps

On 20 October 2026 Google Cloud’s London region had an outage, which Google put down to a power event in one zone, and Revolut card payments were partly unavailable. What a cloud region and zone are, why banks rent the cloud, how one region can stop a card payment, why spreading across regions costs more, and what to do when your bank goes down, with an uptime calculator.

HanaHow15 minute read
3:05 videoWritten steps below

On the morning of Tuesday 20 October 2026, Google Cloud’s London region started having problems, and Revolut card payments stopped working for many customers. People also reported trouble with Lloyds, HSBC and Bank of Scotland, but any link between those and Google is unconfirmed. This guide follows the HanaHow episode: what happened, what the cloud actually is, why one region in London can stop a card payment, and what you can do when your bank goes down.

What happened

Google’s Cloud status page says the problems began at 07:08 UK time. At 09:35 it opened an incident titled “Multiple services are experiencing issues in the europe-west2 region.” europe-west2 is Google’s London region. The incident listed 38 Google Cloud products, among them Compute Engine (its virtual machines), Cloud SQL and Spanner (databases), Cloud Storage, Kubernetes Engine, Pub/Sub and load balancing.

Google’s update at 10:24 UK time said: “Multiple Google Cloud services are experiencing elevated errors and intermittent network failures in europe-west2 region.” Under workaround it said: “If feasible, the impacted customer may try to migrate their workload impacted to another available location (zone or region).” Its earlier update, at 09:35, had said “None for now”.

At 12:12 UK time Google said: “Due to a power event we had suffered an outage within europe-west2-b zone. The power event is mitigated.” europe-west2-b is one of the three London zones, and The Register had named it earlier that morning. Google said services in that zone were unavailable, and that “some services in europe-west2 region may experience elevated errors and intermittent network issues.” Earlier, a Google spokesperson had told LBC: “We are investigating this further.”

Revolut

Revolut said on X: “We are currently experiencing a service disruption affecting multiple services” and “We can confirm that all our customers’ money and data are safe.” Reuters reported that Revolut’s status page showed most of its services “unavailable or partially unavailable, including card payments and ATM withdrawals”, and that “a person familiar with the matter told Reuters that Revolut’s service outage was related to the Google Cloud service disruption.” That link comes from an unnamed source, so it’s reported, not confirmed by Revolut or Google. PA reported more than 1,500 reports about Revolut on Downdetector that morning. When we checked at 12:13 BST, Revolut’s UK status page listed 30 services as partially unavailable, including card payments, transfers and ATM withdrawals, all detected at 08:47 BST.

Other banks: reported, not confirmed

People reported problems on Downdetector with Lloyds Bank (nearly 500 reports by 9:33am), HSBC (over 250 by 10:10am) and Bank of Scotland (about 150 by 9:57am), according to the news site Britbrief. Lloyds Bank told TechRadar: “Our app is online and customers can log in normally, though we understand some are having issues with certain online services this morning.” It didn’t mention Google. Those are customers’ reports, not statements from the banks, and any connection to Google is unconfirmed. Reports citing Downdetector also mentioned Royal Mail, DPD, Rightmove, Screwfix and BT; those aren’t confirmed as linked either. The budgeting app Emma said on X that it was “down due to a Google Cloud outage affecting services in London since 7am”, PA reported.

What a cloud region and zone are

The cloud is other people’s computers: huge buildings full of servers that companies rent by the hour instead of buying their own. On Synergy Research Group’s figures for April to June 2026, Amazon had 28% of the world’s spending on cloud infrastructure, Microsoft 20% and Google 15%. Together the three had 63%.

Google splits its cloud into regions and zones. In its own words, “Regions are independent geographic areas that consist of zones”, and “Zones and regions are logical abstractions of underlying physical resources.” A region is made of three or more zones, usually “housed in three or more physical data centers”, and Google says it “designs zones to minimize the risk of correlated failures caused by physical infrastructure outages like power, cooling, or networking.” So one zone going wrong shouldn’t take the others with it. On 20 October Google said the outage was in one zone, europe-west2-b, but that “some services in europe-west2 region may experience elevated errors and intermittent network issues”.

London, europe-west2, has three zones: europe-west2-a, -b and -c. Google lists all three as “London, England” and doesn’t say which buildings they’re in. It’s Google’s only region in the UK. The nearest other one, europe-west1, is in St. Ghislain, Belgium.

Why banks rent the cloud

A bank’s computers have to cope with its busiest minute: payday, Black Friday, Christmas Eve. Build your own data centre and you pay for that minute all year round. Rent from a cloud company and you can add servers for the busy days and hand them back afterwards, and you don’t have to build or cool the buildings yourself.

Many UK banks rent from Google, among others. Revolut was named as a customer when Google opened its London region in July 2017, and in September 2025 Google announced a multi-year deal to power Revolut’s “rapidly growing global systems”. Google Cloud announced a five-year collaboration with Lloyds Banking Group in March 2020, and HSBC said in June 2026 that “more than 600 HSBC applications” were “already running on Google Cloud”. Banks often use more than one provider: Lloyds was also reported to be hit by Amazon’s outage in October 2025, below. Being a Google customer doesn’t on its own mean a bank was affected today.

Why one region can stop a card payment

When you tap a card, the shop’s card machine sends a request through its payment company and the card network, such as Visa or Mastercard, to the bank that issued your card. The bank checks the card is real, that there’s money or credit to cover it and that it doesn’t look like fraud, then sends back yes or no. All of that has to happen while you stand at the till.

So the bank’s own systems are part of every payment. If the computers that answer for it are in a region with “elevated errors and intermittent network failures”, the answer can come back late or not at all, and the payment usually fails. The same goes for cash machines, transfers and logging in to the app, which all need the bank’s systems to answer.

Zones, regions and why spreading out costs more

Running a service in three zones protects it from one zone failing. Google’s documentation says that if a zone becomes unavailable, “you can transfer traffic to another zone in the same region”. It also says to treat each zone as “a single failure domain”, and that “Putting resources in different regions provides an even higher degree of failure independence.” Its advice: “Google recommends deploying applications across multiple zones and multiple regions.”

For a bank in London, a second region means a second copy of the systems that answer card payments in, say, Belgium, with the data kept in step all the time and ready to take over. That copy has to be paid for every day, for the rare day a whole region fails. Google’s own architecture guide says a setup in one region “is robust against zone outages, but not against region outages”, and that “When a region outage occurs, the application is down.” To recover, “you must have deployed a passive replica of the application in a second (failover) region.” On running in several regions, it says to “consider the potentially higher cost of cloud resources and network traffic, and the complexity of operating the deployment.” Its general guidance puts it simply: “if a single region fails, only customers solely in that region are impacted”.

Some banks build a backup of a different kind. Monzo, which runs on Amazon Web Services, built what AWS’s case study calls “Monzo Stand-in, an independent set of systems that can take over for the primary stack in case of an incident.”

Google’s own service level agreements show what each setup is meant to get you. Checked on 20 October 2026, on Google’s Premium network tier, Compute Engine promises a monthly uptime of 99.9% for a single Compute Engine virtual machine and 99.99% for virtual machines “placed across two or more Zones in the same Region”. Its Spanner database promises 99.99% in one region and 99.999% across more than one. If Google misses one, the customer gets a credit on the bill. It isn’t a promise that nothing goes down.

Uptime calculator

Those percentages sound almost the same. The calculator turns them into time. 99.9% allows 8 hr 45 min 36 sec of down time a year, or 43 min 48 sec a month. 99.99% allows 52 min 34 sec a year, and 99.999% allows 5 min 15 sec.

It also shows what a second copy does, in theory. Two copies at 99.9% that never fail at the same time are both down only 0.1% of 0.1% of the time, which is 99.9999% uptime, or 32 sec a year. That only holds if they fail independently. A bad change pushed to every region at once, or a fault in something every copy relies on, takes them all down together, so treat that figure as the best case.

Uptime and copies

In separate zones or regions, each able to run everything alone

Three nines. What Google promises a month for a single Compute Engine virtual machine. Down 0.1% of the time: about 1 minute in every 1,000.

Down time allowed a year at 99.9%
8 hr 45 min 36 sec
That’s 43 min 48 sec a month or 1 min 26 sec a day. Two copies that never fail together: 32 sec a year, in theory.
Down time allowed for one copy and for independent copies together
Down timeOne copyTwo copies, in theory
Uptime99.9%99.9999%
A year8 hr 45 min 36 sec32 sec
A month43 min 48 sec2.6 sec
A week10 min 5 sec0.6 sec
A day1 min 26 sec86 ms

A year here is 365 days, a month is a twelfth of that (30 days 10 hours) and a week is 7 days. Down time is to the nearest second, or the nearest tenth of a second under 10 seconds. An uptime with more than eight decimal places is shown as a count of nines: 99.9999999999% is 12 nines.

The copies column is a theoretical figure: 1 − (1 − uptime)copies. It only holds if the copies fail independently, never for the same reason at the same time, and if switching from one to another is instant. Real outages often break that: a bad update pushed to every region at once, or a fault in something they all share, takes every copy down together. The Google figures in the notes are the monthly uptimes in its Compute Engine and Spanner service level agreements, checked on 20 October 2026. Missing one earns customers a credit on the bill; it isn't a guarantee the service stays up.

Concentration risk and the critical third parties rules

Regulators call this concentration risk: lots of firms leaning on a few cloud companies, so one bad morning at one of them hits many firms at once. In June 2022 the Treasury said that “As of 2020, for example, over 65% of UK firms used the same four cloud providers for cloud infrastructure services.”

The Financial Services and Markets Act 2023 gave the Treasury the power to name a supplier a critical third party if its failure could “threaten the stability of, or confidence in, the UK financial system”. The Bank of England, the Prudential Regulation Authority (PRA) and the Financial Conduct Authority (FCA) published the rules for those firms in November 2024. They took effect on 1 January 2025 but only apply once a firm is named. They include rules on managing risk, reporting incidents to the regulators and to the banks they serve, and testing for severe disruption. Regulators can publicly censure a critical third party or limit the services it provides to banks; the sections of the Act that cover this don’t provide for fines.

When Amazon’s Virginia region had its outage in October 2025, no firm had yet been named. On 8 July 2026 the Treasury made the Critical Third Parties (Designation) Regulations 2026, naming four from 13 July 2026: Amazon Web Services EMEA SARL, Google Cloud EMEA Limited, Microsoft Ireland Operations Limited and Oracle Corporation UK Limited. The Treasury said the oversight “applies only to the systemic services provided to the financial sector”, and that banks are still responsible for managing their own suppliers. Nikhil Rathi, the FCA’s chief executive, said: “when the same providers serve thousands of firms, a single failure can reverberate across the financial system.”

So Google Cloud EMEA Limited had been under this oversight for just over three months on the morning of the outage.

It has happened before

London, July 2022. On 19 July 2022, the day the UK first recorded 40°C, Google’s London region had “a simultaneous failure of multiple, redundant cooling systems in one of the data centers that hosts the zone europe-west2-a”, according to Google’s incident report. Google powered down part of that zone. It also said that “we inadvertently modified traffic routing for internal services to avoid all three zones in the europe-west2 region, rather than just the impacted europe-west2-a zone”, so services designed to survive one zone failing were hit too. Google said cloud services were restored after 18 hours 23 minutes, and the last effects were cleared after 35 hours 15 minutes. Oracle reported a cooling problem at its London data centre the same day.

Google Cloud worldwide, June 2025. On 12 June 2025, Google said a new feature added to one of its systems, Service Control, “did not have appropriate error handling nor was it feature flag protected.” A policy change with “unintended blank fields” was copied around the world within seconds, and the system crashed in every region. Spreading across regions didn’t help, because the same faulty change reached all of them. Cloudflare said its own outage that day, 2 hours 28 minutes long, came from “a third-party cloud provider”, which it told TechCrunch was Google Cloud.

Amazon’s Virginia region, October 2025. On 20 October 2025, a year before this outage, Amazon Web Services’ us-east-1 region in Northern Virginia had a major outage. AWS said the cause was “a latent race condition in the DynamoDB DNS management system that resulted in an incorrect empty DNS record”, and its summary puts the whole event at about 14 and a half hours. In the UK, LBC and Business Insider reported problems with Lloyds, Halifax and Bank of Scotland online banking and HMRC’s website, and Business Insider reported a Lloyds spokesperson saying issues with AWS were affecting some services.

CrowdStrike, July 2024: not the cloud. On 19 July 2024, a faulty update to CrowdStrike’s security software crashed Windows computers around the world. CrowdStrike said a configuration update “triggered a logic error resulting in a system crash and blue screen”, and Microsoft estimated that 8.5 million Windows devices were affected. It looked like a cloud outage, but it wasn’t one. The fault was in security software on the computers themselves, and CrowdStrike said Linux and Mac systems weren’t affected.

What to do when your bank goes down

  1. Keep a second way to pay. A card from a different bank, or a bit of cash, gets you through a morning like this one.
  2. Check your bank’s own app or status page, or its official social media account, rather than a link someone sends you.
  3. Treat “outage refund” messages as scams. If a text, email or call offers you a refund or compensation for the outage and asks you to click a link, log in or move money, don’t. Contact your bank using the number on the back of your card or in its app.
  4. If you’ve already clicked or paid, call your bank straight away. You can forward a scam text for free to 7726, as the National Cyber Security Centre advises, and report fraud at reportfraud.police.uk or on 0300 123 2040 in England, Wales and Northern Ireland, or to Police Scotland on 101. After the CrowdStrike outage, the NCSC said it had already seen “an increase in phishing referencing this outage”.
  5. Ask about losses. If the outage cost you money, such as a late payment fee or a missed booking, keep evidence and complain to your bank. We couldn’t find any rule that makes a bank pay automatic compensation for an outage, but banks have paid out before: after its outage in January and February 2025, Barclays said it expected to pay £5 million to £7.5 million in compensation, PA reported. If you’re not happy with the bank’s answer, you can take it to the Financial Ombudsman Service.

Where things stand

As of 20 October 2026, about 22:20 BST:

  • Google’s status page now marks the London incident as mitigated, with an end time of 15:00 BST. At 12:12 BST Google said “a power event” had caused “an outage within europe-west2-b zone”, and that the power event was mitigated. At 13:44 BST it said most workloads in the region had recovered. At 15:03 BST it said the problem was mitigated “for most of the customers”, was believed to still affect “a very small number of customers”, and that it would publish an analysis once its internal investigation was complete. It said it would post no further updates on the status page. It hasn’t yet said why one zone caused errors across the region.
  • Revolut hasn’t named Google or said what caused its outage, in the statements we found.
  • Lloyds Bank told TechRadar its app was online and customers could log in normally, though some were “having issues with certain online services”, and that it was “aware of similar issues affecting other organisations”. It didn’t name Google. We found no statement from HSBC or Bank of Scotland.
  • Nothing has been published on whether Google has reported the outage to UK regulators under the critical third parties rules.

More could still come out, and we’ll update this guide when it does.

Sources

This guide is based on what the companies, regulators and news outlets had published by about 13:25 BST on 20 October 2026.

Something not quite right?

Let us know
Google Cloud outage Google Cloud outage today europe-west2 Revolut down Revolut card not working Lloyds down Bank app outage Cloud region Availability zone Concentration risk Critical third parties Uptime calculator Three nines

A little more help

How are computer chips made so small?

How chipmakers fit billions of transistors on a chip: how small a transistor really is and why "2 nm" is only a name, what a transistor does, why shrinking made them leak, the move from flat transistors to fins to gate-all-around, how the floating ribbons are built, how EUV light prints the patterns and how long a wafer spends in the factory. With a size slider from a hair down to the atoms.

Open the guide