At 6:08 this morning, 20 October 2026, part of Google Cloud's London region lost power. Google called it a "power event"; much of the internet called it a very bad Monday. Revolut, Lloyds, HSBC, Royal Mail and DPD were among the services publicly reported down or degraded, and some venues found their tills would not trade through the morning rush. Seamless venues kept serving, because our tills do not need the cloud to take an order. (We build Seamless, so of course we would say that; the sources and the honest caveats are below.)
We are writing this on the day, because today is the one day nobody needs convincing the question matters. By Friday the outage will be a footnote. Your lunchtime queue will still be your lunchtime queue.
What actually happened
A power event took down capacity in Google's London data centre region (europe-west2) from just after 6am UK time. Google reported "elevated errors and intermittent network failures" across multiple cloud services, with recovery running through the morning (Google's incident page, The Register, CNN). Household names across banking, parcels and property went down with it.
Nobody was hacked and nobody was negligent. Clouds fail: power fails, networks fail, whole regions occasionally take the morning off. The best questions are how often, for how long, and what happens to your business while they do.
We heard from venues this morning, not our customers, whose tills would not take orders while the outage ran. We are not naming the systems involved, because we cannot verify every report, and because the point is bigger than any one brand: if your till is essentially a web page, your till is only as available as somebody else's data centre.
Why some tills stopped and some did not
There are two ways to build a cloud EPOS, and this morning sorted them neatly into two piles.
Cloud-tethered tills
The till is a thin screen over someone's server. Every order, every basket, every print job makes a round trip to the cloud. That architecture is cheaper to build and perfectly pleasant to use, right up until the cloud blinks. Then the queue in your shop stops moving, and there is nothing you or your staff can do about it except apologise and point at the sky.
Offline-first tills
The till keeps the menu, prices, orders and tabs in a database on the device itself. Orders are taken and sent to the kitchen locally; receipts print locally. The cloud is for syncing: reporting, menu changes, the back office. When the internet or the cloud disappears, the till carries on trading and catches up when the connection returns. That is how Seamless is built, and it is why this morning our venues' tills did not care what was happening in a data centre in London.
The honest bit
Offline-first is not a magic wand, and we would rather you heard the limits from us than discovered them mid-service.
- Online orders pause. Just Eat, Uber Eats and Deliveroo need the internet to reach you. During an outage, delivery orders queue on the platforms' side, not in your kitchen. Our takeaway EPOS page covers how that works in normal service.
- The back office waits. Reporting and menu edits sync when the connection returns. The till trades; the spreadsheet catches up.
- Card payments are the card machine's business. Your terminal talks to the card networks through your payment provider, not through your EPOS cloud. In this morning's outage the card networks were broadly unaffected, so venues with a working terminal could keep taking payments. Ask your own provider how their terminal behaves offline; answers vary.
- No power, no till. If your electricity is off, no architecture saves you. An iPad on battery with a 4G phone hotspot gets you surprisingly far; a till bolted to mains and ethernet does not.
Five questions to ask any EPOS provider
Ask these in writing. Any provider worth their monthly fee can answer all five without a call back from "a specialist".
- If your data centre goes down at 12:30 on a Saturday, can I take an order at 12:31?
- Can I print to the kitchen with no internet at all?
- What exactly stops working offline, and what catches up by itself afterwards?
- When was your last outage, and do you publish a status page?
- Does "offline mode" mean full trading, or a crippled read-only till?
And one you can answer yourself in two minutes on a quiet Tuesday: turn the Wi-Fi off and try to take an order. If the till stops, you have your answer, and you learned it on a Tuesday rather than during the Saturday rush. Our EPOS comparison is the longer version of that exercise, and the features page covers what Seamless runs locally on the till.
If your till went down this morning
No gloating here; a dead till is a horrible shift. Get your provider's incident report in writing, ask them the five questions above, and if the answers are not reassuring, it costs nothing to look around. Venues that trade through festivals and markets have been solving this for years, which is why food truck operators are the most offline-literate buyers in hospitality: no signal at a festival is just a normal working day.
Frequently asked questions
Does Seamless work without the internet?
Yes. Orders, kitchen printing and tabs run from the till itself, and everything syncs to the back office when the connection returns.
Could venues take card payments during the Google Cloud outage?
Generally yes. Card terminals connect to the card networks through your payment provider, not through your EPOS company's cloud, and the card networks were broadly unaffected on 20 October 2026. Confirm your own terminal's offline and failover behaviour with your payment provider.
What caused the 20 October 2026 outage?
Google reported a "power event" affecting its London (europe-west2) cloud region from 06:08 UK time, causing elevated errors across multiple services through the morning. Google's incident page has the timeline.
How do I know if my EPOS depends on the cloud to trade?
Turn off the Wi-Fi and try to take an order. If the till stops, it is cloud-tethered. Do the test on a quiet day, not during service.
Will this happen again?
To some cloud, somewhere: yes. Regional cloud outages happen every year across every provider. The fix is not avoiding the cloud, it is choosing systems that degrade gracefully when it fails.
Disclosure and sources
We build Seamless, an offline-first EPOS for UK hospitality, so we have an obvious interest in you caring about this. That is why every factual claim above is sourced: Google Cloud's incident page, The Register and CNN, all as reported on 20 October 2026. Reports of specific companies affected come from press coverage and public outage trackers; we have not independently verified them. No provider named here has reviewed this page. Written on 20 October 2026, while the outage was still being resolved.
