Summary

  • Ofcom’s October 2026 explanation separates four jobs: government sets the legal framework, Ofcom supervises compliance, communications providers operate and restore networks, and other sectors—especially electricity—shape important dependencies.
  • Ofcom’s mobile radio access network model estimates that about two-thirds of premises could make an emergency call during an outage of up to roughly one hour. That is not a promise of ordinary voice or data service, and the report does not disclose the modeled six-hour reach.

On 13 October 2026, Ofcom published two pieces that belong together. One tells households and small businesses how to prepare for severe weather. The other explains what the regulator does when communications networks face weather, power, cyber, software or operating failures. The second document is a useful map of responsibility, but its most consequential sentence is not a new guarantee: communications providers remain responsible for designing, operating and restoring their networks and services.

The distinction matters because resilience spans organizations that do not share one control room. Government sets the legislative and policy framework. Ofcom publishes expectations, gathers information about providers’ incidents and dependencies, works with government and expert bodies, and can investigate or enforce when it identifies possible non-compliance. The provider runs the service and carries the operational task of restoration. An electricity network or cloud supplier may be essential to that service without becoming the telecom provider’s substitute decision-maker.

Ofcom’s final Network and Service Resilience Guidance, published on 16 June 2026, makes the provider boundary more concrete. It applies to wholesale and retail communications providers. A provider’s duty includes appropriate and proportionate measures for network portions operated by third parties on its behalf. Buying a wholesale link or outsourcing field operations may change who performs a task; it does not erase the provider’s responsibility for the service it offers. Ofcom also asks providers to understand dependencies across service design, infrastructure, operations, suppliers and transitions.

In a disruption, that makes the provider’s dependency map and operating records as important as a spare battery inventory.

The mobile-power evidence shows why a single resilience percentage can mislead. Ofcom’s February 2025 technical report modeled emergency-call access during power outages. It estimated that, for an outage of up to around one hour, premises representing roughly two-thirds of the population could still make an emergency call. The report linked that outcome to backup power lasting at least an hour at around 20% of mast sites and to handsets’ ability to roam onto an available network for emergency calls.

This is a model of a specific service outcome, not a count of all sites that stay online and not evidence that mobile data or ordinary calls remain available.

Duration changes the result. The report says only around 5% of sites had at least six hours of backup, but it redacts the modeled number of premises that could make an emergency call in that scenario. The missing number is a boundary on what the public can conclude; it is not a blank to fill by extrapolation. Ofcom estimated that upgrades to keep almost everyone able to contact emergency services for up to four hours would cost about £1 billion, including capital and multiple years of operating expense. It explicitly says this estimate is not a cost-benefit analysis and does not assign a value to different resilience levels.

Extending the target to general voice and data would be a different service objective and would cost more.

A separate 2024 Connected Nations report recorded 12 million mobile radio access network customer-hours lost to power issues during September 2023–August 2024. In that reporting snapshot, about 20% of sites had backup to maintain radio access network function for more than 15 minutes, and about 5% could withstand six hours, excluding transmission backup. Those site-level figures and the later emergency-call model have different dates, measures and methods. They should not be read as a time series or as interchangeable evidence. They do show why a regulator needs both infrastructure indicators and service-level outcomes.

For an operator, the useful control question is not merely “How many sites have batteries?” It is: which customers can reach which service, through which surviving path, for how long, and what evidence confirms restoration? A credible operating record would connect power-loss duration and affected geography to live-cell coverage, emergency-call reach, ordinary voice and data availability, customer-hours affected, supplier or electricity coordination, and the time at which each service returned.

The June guidance also points toward tested failover, accurate asset records, service modelling and monitoring that can identify the number of subscribers affected.

The energy dependency makes this an inter-agency problem, but not a reason to blur duties. Ofcom says it works with government, industry and Ofgem to improve understanding and coordination around power. The 2025 report asks how electricity reliability and telecoms backup should be balanced, whether backup should be targeted where long outages are more likely, and how sectors can share information without duplicating investment. Those are real joint-planning questions. They do not make a regulator responsible for dispatching generators or make a power company responsible for restoring a mobile service.

Heng Lu’s Note 70 supplies only an editorial question: continuity of a function is distinct from preserving every institution or authority associated with it. Applied here as a lens, continuity claims should identify the service, the responsible provider and the evidence of restoration. The note is not evidence about Ofcom or UK law.

Ofcom’s severe-weather advice asks customers to charge devices, keep key numbers and consider access to different mobile providers; it notes that emergency calls may use another network when a customer’s own provider is down. Such preparation can reduce harm, especially where rural customers have fewer alternatives. It does not transfer the provider’s statutory or operational duty to households. The practical test for any resilience claim is therefore bounded and demanding: name the service, specify the duration and population, show which dependencies were exercised, and publish the restoration evidence.

A network that can call 999 for one modeled hour has not thereby demonstrated four hours of broadband, an unaffected rural region or a tested recovery plan.

Sources