There is a particular kind of confidence that comes from having solved the wrong problem elegantly. The NHS has spent the better part of the last decade building the case for digital transformation, electronic patient records, AI-assisted diagnostics, federated data platforms, ambient voice tools that transcribe clinical conversations in real time. The vision is coherent, the investment is real, and some of it is beginning to work. But a conversation happening in Silicon Valley right now asks a question that British health policy has largely avoided: what happens when the physical layer underneath all of this starts to give way?

Mike Schroepfer, former chief technology officer of Meta, recently set out the thesis animating his investment firm Gigascale Capital with unusual clarity. As artificial intelligence makes software increasingly cheap to build, he argues, competitive advantage shifts to the hardware, energy systems, and supply chains that underpin everything else. Infrastructure becomes the moat. The physical economy, power grids, transformers, manufacturing, materials, is where the real action is, and it has been systematically neglected.

The NHS did not feature in his remarks. It rarely does in conversations of this kind. But the structural argument maps onto British healthcare with uncomfortable precision.

The health service runs on physical infrastructure that is, in many respects, deteriorating. The NHS estate backlog maintenance figure, the accumulated cost of repairs deferred across hospitals, mental health units, and community facilities, has sat above £10 billion for years and continues to rise. Operating theatres that cannot be used because of ceiling failures, wards closed for remedial works, legacy building stock from the postwar period that was never designed to house the demands of modern clinical practice: these are not marginal problems. They are the physical substrate on which every digital ambition rests.

Energy is a more immediate pressure. NHS trusts are among the largest energy consumers in the country, and the cost shock of recent years hit them hard. Data centres, imaging equipment, ventilation systems, surgical infrastructure, all of it is energy-intensive, and none of it has a credible long-term energy strategy at system level. The point Schroepfer makes about power demand growing faster than supply, and the supply chain for the equipment needed to manage that gap being years behind, is not abstract for a health service trying to run 24-hour acute hospitals on budgets that are already failing to cover the basics.

The medical device supply chain adds a further dimension. The NHS's dependence on global supply chains for consumables, implants, diagnostics, and imaging equipment was exposed during the pandemic and has not been fundamentally addressed since. Where Schroepfer talks about rare earth materials and the fragility of manufacturing dependencies, NHS procurement teams would recognise the concern immediately. The difference is that venture capital firms can move to back domestic alternatives. NHS procurement, governed by frameworks that prioritise price and compliance over resilience, largely cannot.

None of this means the NHS should stop investing in digital tools. The case for electronic patient records, better data infrastructure, and AI-assisted clinical decision support remains strong, and the productivity argument for accelerating that investment is well-founded. But the Schroepfer thesis is a useful corrective to the tendency in health policy to treat digital as a self-contained transformation layer, independent of the physical environment in which it operates.

A trust deploying an AI diagnostic tool in a building with inadequate electrical capacity, unreliable connectivity, and an ageing server room is not at the frontier of healthcare innovation. It is managing a patchwork. The software is only as good as the physical environment that runs it.

The political dimension matters too. The current government has committed to NHS reform and productivity improvement, and technology is central to that argument. But the capital investment required to address the estate backlog, upgrade energy infrastructure, and build resilient supply chains is of a different order to the recurring budget settlements that dominate spending review conversations. It requires long-horizon thinking that British governments, across parties, have consistently found difficult to sustain.

What the emerging consensus in hard-tech investment recognises, that physical infrastructure is not the boring part, but the critical part, is a lesson the NHS has been learning painfully for decades without ever quite applying it. The digital future of the health service will not be built on code alone.