.jpg)
There are troubled technology programmes, and then there are technology programmes whose suppliers are summoned before Congress.
This week, the House Veterans’ Affairs Committee voted unanimously to subpoena Oracle Executive Chairman Larry Ellison and CEO Mike Sicilia after the company declined to attend a hearing examining the Department of Veterans Affairs’ electronic health record modernisation programme. The political temperature is now brutally clear. What began as an attempt to modernise one of the largest healthcare systems in the world has become a test of whether the government can still control the cost, delivery and accountability of enormous digital programmes.
.jpg)
The original ambition was straightforward enough. The VA wanted to replace ageing technology with a modern electronic health record that could better connect veterans’ care across the federal system.
The execution has been anything but straightforward.
The programme began in 2017. The new system was first deployed in 2020, with further deployments in 2022. By 2023, the VA had halted most new implementations after clinicians and veterans reported continuing problems with the system, including concerns around usability, reliability and patient care. The department subsequently entered a programme reset before announcing plans to restart deployments. It now intends to complete approximately 170 sites by 2031.
Meanwhile, the commercial ceiling around the Oracle contract has increased from approximately $10 billion to more than $27 billion.
That does not mean Oracle has automatically been handed a $27 billion cheque. A contract ceiling represents the maximum potential value. But politically, that is almost beside the point. A government technology programme that once came with a $10 billion ceiling can now potentially consume nearly three times that figure.
For lawmakers, the obvious question is simple: how?
.jpg)
The number that should worry healthcare leaders most may actually be 14.
The US Government Accountability Office has issued 18 recommendations across previous reviews of the programme. As of August 2026, only four had been fully implemented. Fourteen remained not fully addressed, including recommendations covering cost estimates, schedules, user adoption, system problems, programme management and operational testing. Twelve of the 18 were considered priority recommendations.
That matters because the VA is now trying to accelerate the rollout.
The GAO’s warning is almost painfully obvious: expanding a programme before some of its most important governance problems have been resolved risks reproducing those problems at scale.
And there is an even more fundamental issue.
The VA still has not produced the full independent life-cycle cost estimate that the GAO has repeatedly requested. It has provided Congress with a notional schedule, but GAO says it has not supplied the detailed supporting documentation necessary to determine whether that schedule follows leading practice.
In other words, the government knows how large the contractual ceiling might become, but still cannot confidently tell taxpayers what the entire programme will ultimately cost.
That is extraordinary.
The project is not merely an enormous future liability.
According to GAO, the VA reported that approximately $13.84 billion had already been obligated to the wider EHR modernisation programme between fiscal year 2018 and the second quarter of fiscal year 2025.
That included approximately $5.85 billion on the EHR contract itself, $3.35 billion on IT infrastructure, $2.85 billion within the Veterans Health Administration, $1.48 billion on programme management and another $324 million through the Office of Information and Technology.
Those numbers matter because they demonstrate one of the central truths of healthcare technology.
The software licence is rarely the real cost.
Infrastructure, implementation, training, workflow redesign, programme teams, clinical change, data migration, integration and support can ultimately become as financially significant as the platform itself.
.jpg)
It would be easy for British healthcare leaders to read this as another spectacularly expensive American government technology story.
That would be a mistake.
The NHS is currently moving deeper into large-scale digital infrastructure, federated data, AI-enabled workflow, shared records, platform procurement and national technology programmes. The parallels do not need to be exact for the lessons to matter.
The first lesson is that scale magnifies weakness.
A problem affecting one hospital is painful. The same problem embedded across dozens of hospitals becomes institutional. Once training, workflows, integrations and local operating models have been rebuilt around a technology platform, reversing course becomes progressively harder and more expensive.
The second lesson is that contract value and programme value are not the same thing.
A supplier may appear to have a tightly controlled contract while the wider cost of implementation explodes elsewhere across estates, infrastructure, workforce, consultancy, integration and local deployment.
That is precisely why full life-cycle costing matters.
The third lesson is that digital transformation cannot be judged by whether the technology technically works.
The real questions are whether clinicians use it effectively, whether productivity improves, whether safety improves, whether patients notice the difference and whether the promised economics eventually appear.
The NHS hardly needs lessons from America about failed technology programmes.
The National Programme for IT remains one of the most famous warnings in British public-sector technology history. It began with ambitions to transform how information moved across the NHS and ended after years of contractual disputes, changing requirements and billions of pounds of expenditure.
The danger now is assuming that modern cloud platforms, AI and better procurement automatically make that history irrelevant.
They do not.
Technology has improved enormously. The organisational problem remains.
Healthcare is extraordinarily difficult to standardise. A digital platform touching thousands of clinicians, dozens of specialties, multiple organisations and millions of patients is not simply an IT implementation. It is organisational surgery.
And surgery conducted without precise visibility over cost, accountability and clinical adoption tends to bleed.
For every major national NHS technology programme, four questions should become unavoidable.
First, what is the full life-cycle cost, not simply the supplier contract?
Second, who is personally accountable for delivering the clinical and financial benefits?
Third, what independent evidence demonstrates that users are actually adopting the technology successfully?
And fourth, what is the exit plan if the supplier underperforms?
Those questions sound basic.
The VA story demonstrates exactly why they are not.
It would also be too simplistic to present this solely as an Oracle failure.
The GAO’s criticism extends directly to the VA’s own programme governance, cost estimating, scheduling, change management and ability to resolve issues. Large government technology failures are rarely caused by one supplier acting in isolation. They tend to emerge from a collision between procurement decisions, political ambition, governance failures, poor programme control and technology that becomes too deeply embedded to replace easily.
That distinction matters enormously for the NHS.
Blaming the vendor after a programme has failed is politically satisfying.
Building a system in which failure is detected early is considerably more useful.
.jpg)
America’s veterans deserve a modern health record. NHS patients deserve modern digital infrastructure too.
But modernisation cannot become a magic word that suspends ordinary financial discipline.
When a healthcare technology programme can move from a $10 billion contract ceiling to $27 billion while its ultimate life-cycle cost remains unresolved and the majority of the auditor’s recommendations remain outstanding, the issue is no longer merely technical.
It is governance.
The NHS is about to spend enormous amounts of money on platforms, AI, digital records, data infrastructure and automation over the next decade.
The lesson from Washington is not that Britain should move more slowly.
It is that Britain should know exactly what it is buying, what success looks like, who is responsible when it goes wrong and how much the entire journey will cost before the train has already left the station.
Because once a national health technology programme becomes too big to fail, taxpayers usually discover something uncomfortable.
It has also become too expensive to stop.
.jpg)
.jpg)
.jpg)