On the morning of 24 June, engineers at Queen Alexandra Hospital in Portsmouth were not managing patient flow or chasing elective recovery targets. They were trying to stop the building from defeating itself. Several chiller units supporting critical infrastructure had failed in the heat, pushing temperatures up across the site and taking down digital systems, operating theatres, cardiac catheter laboratories and diagnostic scanning facilities. The trust declared a critical incident. The following day, Norfolk and Norwich University Hospitals did the same. The trust confirmed it had no working MRI scanners across its Norwich sites, including those at the main hospital and its community diagnostic centre, after cooling systems were overwhelmed by the heat and humidity. At least 254 outpatient appointments were cancelled. Elsewhere, one hospital experienced failures in its laboratories and in two linear accelerator machines used to deliver radiotherapy to cancer patients, while overheating servers prompted staff to switch off non-essential computers and lighting to reduce the strain on systems.

The Met Office issued its first-ever red warning for extreme heat covering three consecutive days, and the NHS, for all the ambition invested in its digital transformation agenda, revealed itself to be partially dependent on cooling infrastructure that cannot reliably function when the outside temperature rises above what has recently become unremarkable for an English summer. That is the quiet scandal behind the headlines about cancelled scans.

The framing of last week's disruption as an emergency response problem, one of triage and resilience planning, obscures the structural failure it represents. The estimated backlog maintenance liability across the NHS estate stood at £15.9 billion in 2024–25, with the total increase in the backlog from 2023–24 to 2024–25 exceeding the total investment made in reducing it. That gap is not a technical anomaly. It is the accumulated consequence of more than a decade in which capital spending on the NHS estate was treated as a residual, available for deferral whenever revenue pressures demanded it. The buildings that now house NHS server rooms, MRI scanners and theatres were, in many cases, not designed with active cooling, not retrofitted as digital infrastructure expanded, and not assessed systematically for climate risk. They were simply used, maintained at the margins, and asked to carry ever-increasing clinical and technological load.

The government's response to the structural problem has been to announce it. The 10-year infrastructure strategy commits at least £64 billion over the next decade to maintain the health estate, with a dedicated Estates Safety Fund of £750 million a year over nine years targeted at reducing critical infrastructure risk. That commitment is meaningful in scale. It is also timed in a way that should concern anyone who observed what happened to Norfolk and Norwich last week. The summers are not waiting for the programme to complete.

There is a particular irony in the collision between infrastructure failure and digital ambition. The government's ten-year health plan aspires to make the NHS the most AI-enabled health system globally, with hospitals fully AI-enabled within the plan period and a dedicated technology and productivity ringfence of around £1 billion a year to drive it. Running AI at scale in clinical environments generates heat. It requires consistent power. It depends on cooling systems that are demonstrably not up to the job in the trusts currently under the most pressure. Investing in digital capability while deferring the physical infrastructure that supports it is not a sequencing strategy. It is a contradiction.

MRI scanners are used to diagnose strokes, cancers and other serious illnesses. Radiotherapy equipment is essential for many patients receiving scheduled cancer treatment. IT outages can delay access to medical records, diagnostic imaging and laboratory results. These are not peripheral services. When they fail because a building cannot stay cool, the waiting list does not pause. The patient whose scan was booked for Thursday remains undiagnosed. The treatment pathway that depends on that scan does not reset.

NHS leaders and integrated care boards planning capital allocations should draw a specific lesson from last week. Climate resilience is not a net-zero commitment to be discharged through a carbon reporting framework. It is an operational risk with immediate consequences for patient safety and service delivery. Cooling infrastructure, server room design, and the climate tolerance of diagnostic equipment need to appear in trust risk registers not as long-term considerations, but as present ones. The Estates Safety Fund, if it is to justify its name, needs to prioritise heat resilience alongside fire safety and structural compliance.

One senior clinician described the sense of foreboding that comes with watching the weather forecast, and the limited options available when what follows arrives. That foreboding is well-founded. The question is whether the institutions responsible for NHS infrastructure will act on it before next summer's forecast, or wait for another round of cancelled appointments to make the argument again.