Find the Right GRC Solution

Search and compare GRC technology built for the problems you’re trying to solve

Too Critical to Cut Off

Too Critical to Cut Off

By
🔒
Cancel anytime
Full archive access
Custom alerts
Key Takeaways
  • The Most Disruptive Remedies Are the Most Constrained: Both regimes' heaviest step separates financial firms from the provider, and both condition it on the harm that separation would cause. FSMA section 312R(4) requires the regulator to be satisfied that a prohibition, condition or limitation will not threaten UK financial stability or confidence, a test that weighs most on orders to stop using an existing service, and DORA Article 42 requires authorities to weigh business-continuity risk before ordering suspension or termination.
  • The Toolkits Differ in Kind: The UK regime pairs binding directions under section 312N, which are not subject to the section 312R(4) stability condition, with censure and graduated restrictions, but it contains no power to fine. DORA's periodic penalty payments of up to 1 percent of average daily worldwide turnover compel cooperation with oversight, while its recommendations are backed by public disclosure and follow-up through financial entities' own supervisors.
  • Substitutability Is Shaped Elsewhere: The levers that bear most directly on portability sit with competition and data law. The CMA accepted voluntary egress and interoperability steps from AWS and Microsoft in March 2026 and has provisionally proposed designating Microsoft, and the EU Data Act bars cloud switching charges from January 12, 2027. Cheaper exit still does not make a provider-specific architecture portable in an emergency.
  • Designation Follows Entities, Failures Follow Architecture: The UK and EU designated different Oracle entities, and the October 2025 AWS outage that disrupted UK banks began in a US region. Group-wide information powers and the EU-UK memorandum of January 14, 2026 narrow that gap without closing it.
  • Firms Shape the Supervisor's Options: Because the most disruptive sanctions turn on whether customers could continue without the provider, an institution's tested ability to run important business services elsewhere or in degraded form gives supervisors more room to act.
Deep Dive

On the morning of October 20, 2025, customers of Lloyds Bank, Halifax and Bank of Scotland found they could not get into their mobile banking. Lloyds told customers that problems at Amazon Web Services were affecting some of its services. The trouble had begun a few minutes before 8 a.m. London time, several thousand miles away in AWS's Northern Virginia region. According to AWS's own post-event summary, a latent race condition in DynamoDB's automated DNS management left the service's regional endpoint with an empty DNS record, so systems that depended on DynamoDB there could no longer reach it. The disruption spread to other AWS services and did not fully end until late that evening, UK time.

Nine months later, on July 10, 2026, HM Treasury named Amazon Web Services EMEA SARL, a Luxembourg company, as one of Britain's first four critical third parties. The others were Google Cloud EMEA Limited, Microsoft Ireland Operations Limited and Oracle Corporation UK Limited. The designations, made by statutory instrument, took effect on Monday, July 13, and since then the Bank of England, the Prudential Regulation Authority and the Financial Conduct Authority have jointly overseen the services those companies provide to UK financial firms.

The European Union moved first. On November 18, 2025, the European Banking Authority, the European Insurance and Occupational Pensions Authority and the European Securities and Markets Authority, acting together as the European Supervisory Authorities, designated 19 critical ICT third-party providers under the Digital Operational Resilience Act. The list included the hyperscalers' European entities alongside data-center operators, network carriers, consultancies and financial-data firms such as Bloomberg and LSEG Data and Risk.

Much of the commentary has treated the two regimes as the moment financial supervision caught up with its own infrastructure. That is fair as far as it goes. A closer reading of the statutes reveals a more specific difficulty. Both regimes give supervisors genuine access to the providers and real instruments for compelling change. Their most disruptive remedies, however, work by separating financial institutions from the provider, and both legal texts direct the authority to weigh the damage that separation would cause. The more completely a provider's customers depend on it, the harder those remedies become to use. Supervisors keep meaningful leverage. It lies mainly in instruments that change how a provider operates, and it thins out at the point where the remaining step would be to take the provider away.

Two Toolkits, Unevenly Sharpened

The designation tests are close cousins. Section 312L of the Financial Services and Markets Act 2000, inserted by the Financial Services and Markets Act 2023, lets the Treasury designate a provider when it considers that a failure or disruption of its services could threaten the stability of, or confidence in, the UK financial system, having regard to the materiality of the services and the number and type of firms that use them. DORA's Article 31 sets four criteria. They are the systemic impact of a disruption, the systemic importance of the financial entities that depend on the provider, their reliance on it for critical or important functions, and the degree to which its services can be substituted. The European authorities built their first assessment from the registers of information that financial entities filed with national supervisors, which gave Europe's regulators their first sector-wide view of third-party dependencies rather than one firm's contracts at a time.

The powers that follow designation diverge more than the announcements suggested. British regulators can make rules binding on designated providers. Those rules took legal effect on January 1, 2025, and bound no one until July. The regulators can gather information, including from persons connected with a designated provider, commission skilled-person reviews, and issue directions under section 312N requiring a provider to do, or refrain from doing, something specified. Providers must carry out scenario testing, report incidents and submit regular self-assessments. What Chapter 3C of the act does not contain is any power to impose a financial penalty. Its sanctions are public censure under section 312Q and, under section 312R, a notice that prohibits or places conditions on the provider's services to financial firms, or on the firms' receipt of them.

DORA assigns each designated provider a lead overseer drawn from the three authorities, chosen according to which sector's customers hold the largest share of assets, and supported by joint examination teams. The lead overseer can demand information, open general investigations and conduct on-site inspections. It can also impose periodic penalty payments of up to 1 percent of the provider's average daily worldwide turnover, accruing daily for as long as six months. Those payments are designed to compel cooperation with the oversight process itself, meaning information requests, investigations and inspections. They do not attach to oversight's substantive judgments, which take the form of recommendations.

A provider has 60 days to say it will follow a recommendation or to explain why it will not, and the follow-through is more structured than that choice suggests. Under Commission Delegated Regulation (EU) 2025/295, a provider must give the lead overseer a remediation plan setting out the actions it intends to take within the timelines attached to each recommendation, and must supply interim progress reports and final reports on request so the lead overseer can monitor implementation. National supervisors, for their part, assess what the provider's remedies mean for the financial entities under their remit. Recommendations are not enforced through penalty payments, but a provider cannot ignore them at no cost. An absent or insufficient explanation leads to public disclosure naming the provider, the recommendations go to the national supervisors of the financial entities that use it, and those entities must take the identified risks into account in managing their own third-party risk. Designated providers based outside the EU must also establish a subsidiary in the Union within 12 months.

The result is an asymmetry that is worth stating precisely. Britain can issue binding directions and cannot fine. Europe can fine, though only to secure cooperation with oversight, and its conclusions about what a provider should change carry force indirectly, through public disclosure, through customers who have been formally told of the risks, and ultimately through those customers' own supervisors.

The Limits of the Last Resort

Short of any last resort, the British toolkit is broader than its sanctions provisions suggest. A regulator may give a direction under section 312N whenever it considers this necessary or expedient to advance its objectives. Section 312O ordinarily requires written notice setting out the direction and the reasons for it, with an opportunity to make representations, and by default a direction takes effect only once the matter is no longer open to review. Where the regulator reasonably considers it necessary, however, a direction can take effect immediately or on a specified date, with representations considered afterward and a right to refer the matter to the Upper Tribunal. Directions are not subject to the specific financial-stability condition that governs prohibition notices. Where a direction is given to resolve or reduce a threat to the stability or integrity of the UK financial system, the provider and its staff are immune from damages for acting in accordance with it. Directions are the instrument a UK regulator would use to require a provider to change how it runs a critical service, and they work without separating anyone from anything.

The difficulty sits further along the escalation path. Section 312R lets a regulator restrict a critical third party's relationship with financial firms in graduated ways. A notice can bar new arrangements, attach conditions or limitations to particular services, or, at the extreme, prohibit firms from continuing to receive the provider's services. Subsection (4) allows any of these only if the regulator is satisfied that the step is appropriate, advances one of its objectives and will not threaten the stability of, or confidence in, the UK financial system. That condition is easily met by a notice that limits new business or imposes conditions. It weighs most heavily on the most disruptive form, an order that firms stop using a service they already rely on. A provider is designated because its disruption could threaten stability. The more completely the system depends on it, the harder it becomes to show that forcing firms off it would not.

DORA reaches a similar place by a different road. The lead overseer cannot impose penalty payments for a declined recommendation. Article 42 instead pairs public disclosure with follow-up through national competent authorities, which must inform their financial entities of the risks identified. Where a financial entity fails to address those risks, its supervisor may notify it that a decision could follow and, as a last resort, require it to suspend use of the service in whole or in part, or to terminate the contract. Before doing so, the authority must consider, among other factors, whether suspension or termination would create a risk to the financial entity's business continuity, and it must give the entity enough time to adjust its contracts and carry out its exit and transition plans. The heaviest lever is pulled on the customer rather than the provider, and the regulation instructs whoever pulls it to check first whether the customer can survive the pull.

None of this leaves supervisors without leverage. Directions, conditions on new business, public censure and public disclosure can all be used against even the largest providers, and each carries weight for companies whose business with regulated industries depends on being trusted. A censure under section 312Q or a disclosure under Article 42 would be read in every customer's boardroom. What the drafting shows is narrower. The interventions that would reduce the system's exposure to a provider by moving firms off it are the ones most constrained by that exposure. Supervisors can press a hyperscaler to run its critical services better. They are far less able to make the financial system need it less.

Where Substitutability Is Actually Decided

If financial supervisors cannot readily penalize non-substitutability, the obvious question is who can reduce it. Their statutory concern is the resilience of the services. The forces that determine whether a bank could move a workload, among them egress charges, software licensing terms and the interoperability of proprietary managed services, fall largely to other authorities.

In Britain, the Competition and Markets Authority's cloud services market investigation reported in July 2025 that AWS and Microsoft held significant market power, each with up to 40 percent of UK cloud infrastructure services by value, and recommended that the authority consider designating both with strategic market status. On March 31, 2026, the CMA took a different course for cloud. It accepted voluntary steps from both companies, among them free data transfer for UK customers switching away, at-cost transfers for multi-cloud use and direct connections between their data centers and those of rivals.

Some of the egress terms are now written into UK contracts, and the CMA says it retains the option of formal intervention. It also opened a strategic market status investigation into Microsoft's business software ecosystem, which began in May, and on October 16 it published a provisional decision proposing to designate Microsoft. Designation would allow the CMA to act on the software licensing practices its cloud investigation identified as limiting customer choice. The CMA's board plans to review progress on cloud in spring 2027. In the EU, the Data Act's cloud-switching rules have applied since September 12, 2025, and from January 12, 2027, Article 29 bars providers from imposing switching charges, which the act defines to include egress charges for the switch. The Commission has since proposed amendments to the Data Act in its Digital Omnibus package, so the final shape of some provisions remains open.

These measures bear on substitutability more directly than anything in DORA's oversight chapter or the UK's Chapter 3C. They also have limits that an operational resilience specialist will recognize immediately. Free egress makes leaving cheaper. It does not make an application built on one provider's serverless functions, identity services and managed databases run anywhere else. Lower switching costs shorten a planned migration measured in quarters and do little for an emergency measured in hours.

What emerges is a division of labor that neither regime describes. Financial supervisors examine the provider's controls. Competition and data authorities shape the terms of exit. The capability that would matter most in a severe disruption, a tested ability to keep important business services running elsewhere or in a deliberately degraded form, remains with each financial institution, which is precisely where both regimes say it stays.

Legal Entities and Failure Domains

The October outage points to a second, quieter mismatch. Designation attaches to legal entities. Britain designated Oracle Corporation UK Limited, while the EU designated Oracle through Oracle Nederland B.V., so the two systems are supervising different corporate faces of the same provider. Failure domains, by contrast, follow the architecture. An incident that began in a US region reached British retail banking customers within hours.

The statutes anticipate some of this. UK information-gathering powers extend to persons connected with a designated provider, including group companies, and DORA requires designated providers to report their subcontracting arrangements to the lead overseer. The supervisory perimeter is nonetheless a map of contracting entities laid over infrastructure that was never built to respect it.

The memorandum of understanding signed on January 14, 2026, by the three European authorities and the Bank of England, the PRA and the FCA is the formal attempt to bridge the jurisdictional part of that gap. It sets out principles and procedures for cooperation, information exchange and the coordination of oversight activities, and it was preceded by a European assessment confirming that the UK's confidentiality and professional secrecy regime is equivalent to DORA's, a legal precondition for sharing supervisory information with a non-EU authority. It creates no joint supervisor, and each authority retains its own powers. Its practical worth will be tested during an incident affecting both markets, when two sets of supervisors need the same facts from the same provider at the same moment. Coordination can align what the authorities know. It cannot conjure capacity that does not exist.

What the Regimes Buy, and What They Leave to Firms

It would be a misreading to conclude that the regimes are hollow. Supervisors who once saw a hyperscaler only through the outsourcing files of individual banks can now inspect it, examine its incident response and compare what it tells them with what dozens of its customers report. They can require testing against severe but plausible scenarios, and the UK authorities have made timely, open communication with regulators and customer firms during major incidents an explicit expectation. These gains bear on the likelihood of a failure and the quality of the response to one.

The regimes also accept, rather than undo, the decisions that produced the concentration. Financial institutions went to the hyperscalers for scale, security engineering and managed services they could not economically build themselves, and multicloud designs frequently add cost, complexity and new failure modes while delivering less redundancy than their architecture diagrams imply. Neither DORA nor the UK regime requires institutions to spread workloads across providers.

DORA's Article 29 obliges financial entities to assess concentration risk before contracting, including whether a provider is not easily substitutable, and the regulation surrounds that assessment with requirements on contractual terms and on exit strategies, including transition periods, for services supporting critical or important functions. Those provisions make dependence more visible and departure better prepared. None of them requires a firm to reduce its reliance on a particular provider.

The practical consequence is sharper than the familiar warning that concentration risk persists. Designation gives customers some new information, though considerably less than it gives regulators. In Britain, a critical third party must share its annual self-assessment with the firms it serves, redacted as appropriate, while skilled-person reports and the regulators' own findings stay with the regulators. In the EU, firms learn of the risks identified in the lead overseer's recommendations through their national supervisors rather than seeing the underlying examination work. Firms will still have to build their own picture of a provider's resilience. Nor does designation change the question that ultimately falls to the firm, because both regimes route their most disruptive decisions back through it.

When a UK regulator weighs a prohibition notice, or an EU authority weighs a suspension under Article 42, a central fact will be whether the affected institutions could keep their important business services running without the provider in question. A firm that can demonstrate that capability, through tested failover, rehearsed degraded-mode operation or a genuinely executable alternative, gives its supervisor more room to act. A firm that cannot leaves the regulator confronting the very dependency it was appointed to oversee.

 The GRC Report is your premier destination for the latest in governance, risk, and compliance news. As your reliable source for comprehensive coverage, we ensure you stay informed and ready to navigate the dynamic landscape of GRC. Beyond being a news source, the GRC Report represents a thriving community of professionals who, like you, are dedicated to GRC excellence. Explore our insightful articles and breaking news, and actively participate in the conversation to enhance your GRC journey.

Oops! Something went wrong