On Tuesday 20 October, at 8:08 (Swiss time), the Google Cloud status page reported that “multiple services are experiencing issues in the europe-west2 region”, in other words London. During the morning, Revolut showed most of its services as unavailable or degraded, starting with card payments and cash machine withdrawals. Here is what has been established, what rests on statements and what remains open.

What Google Cloud has acknowledged

Google refers to “elevated errors and intermittent network outages”. According to Journal du Coin, the list of affected products grew over the course of the morning to 31 entries, including Compute Engine, Kubernetes Engine, Cloud Storage, Cloud SQL and BigQuery. Put simply: virtual machines, databases and storage, the very building blocks on which a banking app runs.

At 11:24, Google had still not put forward any cause. Its engineers “continue to investigate”, and the only advice given to customers was to move their workloads to another zone “if possible”. A further update was promised for 1 p.m. (13h). No expected duration has been communicated.

The London region has suffered an incident before. In August 2024, the simultaneous loss of both power supplies at a point of presence caused two hours of latency on load balancers across the entire region. For now, nothing suggests that the cause is the same this time.

Revolut: what the neobank says

On X, Revolut confirmed a “service disruption affecting several services” and stated that “all our customers’ money and data are safe”. Downdetector counted more than 1,500 reports shortly after 9 a.m.

The link with Google Cloud rests, at this stage, on a source close to the matter cited by Reuters, according to whom the Revolut outage does indeed stem from the incident. The neobank has not confirmed this publicly, and it has given no time for service to be restored.

Other names appeared on Downdetector between 8 and 11 a.m.: Lloyds, HSBC, Royal Mail, Rightmove, DPD and GOV.UK. Lloyds said its app remained accessible, with a spokesperson referring to “similar issues at other organisations”, without mentioning Google. HSBC, NatWest, Barclays, Santander, Nationwide and Monzo appeared on the same list without having commented by midday. In Germany, ING reported login problems with its online banking, with no established link to Google.

And for customers in Switzerland?

The available information does not detail the situation country by country. If the Revolut account is run on the same infrastructure as the one affected, the effects described apply to everyone: card declined at the till, withdrawals impossible at the cash machine, app inaccessible. Revolut has more than 50 million customers according to our source; nothing suggests that the outage is limited to the United Kingdom, but its extent outside London is not documented.

One point concerns those who have bought cryptocurrencies in the app. These are held by Revolut, not in a wallet whose keys the user controls. As long as the app is unavailable, the sell and withdrawal functions are too. According to our source, the assets are not at risk: it is access that is cut off, not the balance.

In practical terms, the lesson is above all a practical one: a single card, tied to a single provider, is not enough on the day of an outage. Having a second means of payment with another institution limits the inconvenience.

Cloud resilience: the underlying question

According to Journal du Coin, three players, Amazon, Microsoft and Google, host nearly two-thirds of the world’s cloud, and banks and fintechs alike have moved their core applications there. The incident comes exactly one year to the day after the AWS outage that took Coinbase offline, and after the Cloudflare outage last November, which cut off Coinbase and Kraken for several hours.

Our analysis: Google’s advice, to switch “if possible” to another zone, sums up the problem. Cloud resilience does not come automatically with the provider; it depends on how each customer has distributed its services across regions, or even across providers. An application designed to run in a single region stops when that region does. Redundancy has a cost, and it is up to each bank to decide how far it goes. In Switzerland too, the infrastructure choices of banking platforms are a matter of debate, as shown by Swisscom’s use of Oracle Exadata Cloud@Customer for Finnova.

Finally, the source points out that the Bitcoin network continued to produce blocks during the morning, its validation being spread across more than 20,000 nodes. The contrast has its limits, it acknowledges: exchanges and consumer wallets also run on AWS or behind Cloudflare.

The next pieces of information awaited are Google’s explanation of the cause and a restoration time from Revolut. Neither was known at midday.

Sources

  • Journal du Coin: « Panne Google Cloud : Revolut paralysé, des millions de cartes bloquées »