Somewhere in San Francisco this week, engineers at a seven-year-old company called Candid Health were toasting a funding round that tripled their valuation inside eighteen months. Sixth Street Growth led the $120 million Series D, joining investors who have already poured hundreds of millions into a platform built to do one thing: automate the maddening business of getting American hospitals paid. Candid's pitch is that clinical records, billing codes and provider data can be fused into a single model, letting AI agents chase claims, catch errors and close the gaps that cost the US healthcare system tens of billions a year in wasted administration. Annual recurring revenue is reportedly up 190 per cent year on year. Investors are not betting on a niche product. They are betting on a market failure so large and so lucrative that automating it has become one of the more reliable trades in American health-tech.

None of this translates directly to the NHS, and it would be a mistake to pretend otherwise. Britain does not run a per-patient reimbursement system riddled with insurer disputes and coding arbitrage, so there is no equivalent commercial prize waiting to be captured by whoever builds the smartest billing agent. The NHS's problem is not that providers cannot get paid. It is that the state has spent years trying, and largely failing, to build the very thing Candid is selling: a unified data layer that lets an institution see its own patients, costs and operations clearly enough to act on them.

That failure has a name, or rather two. The Federated Data Platform, built with Palantir at a cost that has repeatedly strained political tolerance, was meant to give NHS trusts something close to what Candid now offers American hospitals off the shelf. Years on, the programme remains contested, its governance scrutinised, its benefits harder to demonstrate than its critics' objections. The contrast is instructive. American health-tech investment can move at the speed of a venture round because there is a direct commercial return sitting at the end of the automation, insurers pay faster, denials fall, cash flow improves. British attempts at the same underlying capability have to be justified, procured and defended through a political process that treats every large data contract as a potential scandal in waiting, particularly when the vendor's name carries the associations Palantir's does in Britain.

This is the structural point worth sitting with. The gap between Candid's funding round and anything comparable in the UK is not really a gap in ambition or technical understanding among NHS leaders, many of whom watch American health-tech closely and know exactly what agentic automation could do for referral management, discharge planning or the endless administrative load that currently falls on clinical staff. The gap is that Britain has no market mechanism generating the kind of commercial incentive that built Candid. Instead it has a single payer, a Treasury settlement, and a public that reacts to the words "data" and "AI" in healthcare with understandable suspicion born of past failures.

That leaves NHS leaders and the life sciences sector with an uncomfortable choice. Either the state finds a way to fund and defend data infrastructure on its own terms, accepting the political cost that comes with it, or it waits for adapted versions of American platforms to arrive, stripped of the billing logic that does not apply here but retaining the automation layer that might. Neither path is straightforward. But the alternative, another decade of fragmented records and manual administrative work absorbing clinical time that should go to patients, is the one Britain can least afford. Candid's investors did not fund a vision. They funded a return. The NHS has yet to work out how to make anyone want to fund its equivalent.