Revolut’s lost hour on 20 October is a due-diligence question

07:08 UK time. At that point Google Cloud’s London region, europe-west2, started reporting elevated errors and intermittent network failures across 36 listed products.¹² Revolut went down at roughly 08:00, and the hour in between interests me more than the outage itself. Revolut published no exact timestamp of its own, so the hour rests on press reporting.³ When the outage came, it was broad: cards, transfers, ATM withdrawals, trading and sign-in all failed or degraded, with reports from the UK and Ireland to France, Spain, Romania and Germany.⁴⁵¹

Stat ~1 hour Google europe-west2 errors from 07:08 UK; Revolut down from about 08:00 UK ¹¹

A cascade that takes an hour to arrive says something about the plumbing. Reading the reports, I first asked what a credit committee would ask: what caused it, who else was hit, and what did it cost.

Google has published no root cause.⁶⁷ The link to Revolut rests on one person familiar with the matter speaking to Reuters.⁸⁹ Reporters listed Lloyds, HSBC, DPD and Rightmove among those hit, though nothing I read shows that every disruption shared one cause.⁵¹⁰ I found no public count of affected Revolut customers beyond more than 1,500 problem reports on Downdetector, which is a count of complaints, not of customers.¹¹¹² The cost is blank too.

The case for a bad Tuesday

This was a time-bounded third-party outage, and nothing public suggests that Revolut’s banking systems or asset controls failed.¹³ Revolut said customers’ money and data were safe, and reports placed large incumbents down the same morning.¹³⁵⁵ At this size, one hyperscaler region can also be the rational choice: one contract to negotiate and audit, one environment for engineers to know deeply. I have seen no published source make that case, so it stands here as argument.

I concede it on controls, and on efficiency. But one contract and one environment also make one failure domain, and the hour is where that cost shows. An outage that hits everyone makes the risk shared, not absent.

What the record does not say

Revolut and Google Cloud expanded a multiyear agreement in September 2025, including Google’s AI tools for fraud detection.¹ I found no 2026 on-the-record statement from Revolut’s leadership on listing timing or venue. Nor did any disclosure I read name Google Cloud as a material outsourcer. The last blank weighs most: nothing I found describes a failover architecture.

My reading follows. If and when Revolut lists, the hour is what an investor will want explained, because it is the one public trace of whether a second route existed. That holds only if the reported link between the two events holds, and that link rests on a single unnamed source.⁸⁹ Even then, the hour raises the question and settles nothing. Scale does not dilute it: the more business runs on one region, the more weight rests on the same foundation.

I suspect the same missing page sits in many mid-market counterparty files. A deal runs on platforms the sponsor rarely audits: the loan servicer, the payments processor, the KYC provider. The question I would put to each is what happens in the first hour after its cloud region starts throwing errors, and where that is written down. For Revolut, the answer would begin at 07:08.

This note observes public reporting. It is not investment advice and not a recommendation to buy, sell or hold any security.