25th August 2026
For decades, successive British governments have been attracted to the same proposition to let private companies do the work and the taxpayer will save money.
The theory sounds convincing as a private company can supposedly operate more efficiently, introduce new technology, employ specialist staff and deliver services at a lower cost than the state. Competition between suppliers is supposed to keep prices down and encourage better performance.
But there is an awkward question that governments have too often failed to ask.
What happens when the private contractor gets it wrong?
The answer is becoming increasingly familiar. The Government cannot simply walk away.
The service is too important to stop. The contractor has to be paid, monitored, investigated or penalised. Eventually another contractor may have to be brought in — or the Government may have to take the work back itself.
And by then, the supposed saving can have disappeared.
The latest crisis surrounding Capita's administration of Civil Service pensions provides a particularly uncomfortable example. Capita took over the £239 million Civil Service pension administration contract in December 2025. At the time of handover there were around 85,000 outstanding cases. That backlog subsequently rose to about 120,000, including retirement, bereavement and pension quotation cases. The Government has withheld £9.9 million in payments because of missed contractual deadlines and is now considering whether the service should be brought back in-house.
This is not simply an argument about computer systems or contractual performance.
Behind every unresolved pension case is a person who may have spent decades working for the state and who is entitled to receive money that should be available when they retire.
The Government has described the service levels as "completely unacceptable".
That raises a very basic question.
If the Government has to employ people to monitor the contractor, investigate failures, withhold payments, manage the recovery programme and eventually consider taking the service back in-house, how much of the original saving is actually left?
The Capita experience isn't new
This is important because the current pension problem isn't an isolated incident.
The National Audit Office examined Capita's NHS primary-care support contract several years ago and reached remarkably similar conclusions.
NHS England had awarded Capita a seven-year, £330 million contract in 2015, with an objective of reducing costs by 35%. But the NAO found that neither NHS England nor Capita had fully understood the complexity and variation of the services being outsourced.
The consequences included delays in transferring medical records and problems with support services. Five of the nine services were placed into a formal improvement process, and the NAO said the failures had the potential to cause serious harm to patients.
The NAO's conclusion was particularly revealing.
It recommended that NHS England should consider whether the services were actually better delivered through the contract or whether some should be taken back in-house.
That was eight years ago.
Yet Britain is still having essentially the same argument.
Fujitsu is an even more uncomfortable example
Then there is Fujitsu and the Post Office Horizon scandal.
Horizon was not simply an unsuccessful IT project. Faults in Fujitsu's system contributed to one of the greatest miscarriages of justice in modern British history, with sub-postmasters wrongly prosecuted, convicted and financially ruined.
By March 2026, more than 11,300 claimants had received redress totalling £1.44 billion, while Parliament's Business and Trade Committee was still warning of "serious structural failings" in the compensation system. The committee also highlighted the extraordinary fact that Fujitsu had contributed nothing to the compensation bill while continuing to expand its public-sector business.
That should cause governments to think very carefully.
It isn't necessarily an argument that Fujitsu should never again be allowed to work for government.
But it certainly raises the question of whether a company that has been so deeply associated with a catastrophic public-sector technology failure should automatically remain an important supplier to the state.
More importantly, it raises the question of where the Government's own technical expertise had gone.
If a government department cannot independently understand, interrogate and challenge a system on which people's livelihoods depend, it has outsourced more than a computer system.
It has outsourced part of its ability to govern.
And then there is Palantir
Palantir represents a slightly different problem.
Its technology is being used extensively within the NHS Federated Data Platform, and the system has attracted support because it can bring together complicated NHS information and potentially improve the management of patients, waiting lists, operating theatres and hospital capacity.
So this is not a case where it would be fair to say simply that Palantir has "failed".
There are claims of operational benefits, and the NHS says the platform is already being used widely.
But the issue is becoming increasingly important for a different reason: strategic dependence.
A Parliamentary committee warned in June that Palantir's growing role in the UK public sector represented an "unacceptable point of weakness" and urged the Government to use the 2027 break clause in the NHS contract, either developing an in-house replacement or seeking a UK alternative.
The issue is particularly significant because the Government's latest estimates put the lifetime cost of the NHS Federated Data Platform at approximately £1.1 billion, while projected financial benefits have fallen to around £808 million.
That doesn't automatically mean the system is poor value.
Some of its benefits are difficult to measure in pounds and pence.
But it does demonstrate why government needs to retain sufficient technical expertise to understand what it is buying and to be capable of replacing it if necessary.
Greater Manchester provides an interesting counter-example. It has resisted full adoption of Palantir's system and continues to use its own in-house data platform. Its argument is that its own capability is more flexible and, in some respects, ahead of the national system.
That is an important lesson.
Government can sometimes build and maintain highly capable systems itself.
The assumption that the private sector will always do it better is simply not supported by every example.
The problem with outsourcing isn't private enterprise itself
This is where I think the debate needs to become more sophisticated.
There is nothing inherently wrong with government employing private companies.
Private firms can be extremely good at what they do.
They can provide specialist engineering, construction, software development, consultancy and other services that would be wasteful for government to maintain permanently.
The problem occurs when government starts outsourcing core institutional knowledge.
If a department loses the people who understand how its systems work, how its data is structured, how its processes operate and how suppliers should be challenged, it becomes dependent on the contractors themselves.
And that creates an extraordinary imbalance.
The contractor knows more about the system than the Government that is paying for it.
At that point, the Government is no longer really managing the contractor.
It is managing its dependence upon the contractor.
The cheapest bid isn't necessarily the cheapest solution
This is perhaps one of the biggest lessons Britain needs to learn.
A contract can appear cheaper on paper because the initial price is lower.
But the true cost is much larger.
Government must include the cost of procurement, contract management, monitoring, auditing, legal advice, renegotiation, data migration, transition arrangements and contingency planning.
Then there is the cost of failure.
If a private contractor makes a mistake in an ordinary commercial business, the customer can potentially take its business elsewhere.
Government doesn't have that luxury.
A failed tax system cannot simply be switched off.
A failed pension system cannot tell pensioners to wait.
A failed NHS data system cannot simply be abandoned if millions of records have become dependent upon it.
A failed defence system cannot be allowed to fail because the contract has expired.
The state remains responsible.
That means the taxpayer ultimately carries a significant proportion of the risk even when the service has been privatised or outsourced.
There is also a problem with "temporary" outsourcing
One of the most damaging aspects can be the loss of institutional memory.
Suppose government has a department employing 500 people who understand a complicated service.
A decision is then made to outsource it.
Many of those employees leave.
The contractor takes over.
Ten years later, the Government decides the contract isn't working.
It wants to bring the service back.
But where are the people who know how it works?
They have retired, moved elsewhere or been made redundant.
The Government may therefore discover that bringing something back in-house is much more difficult than outsourcing it in the first place.
It has lost the very capability it now needs.
That is a serious strategic weakness.
What should be brought back?
I wouldn't advocate nationalising everything.
That would simply replace one ideology with another.
Instead, I would argue for a much more rigorous test.
Before outsourcing a public service, government should ask whether it involves core public knowledge, sensitive data, critical national infrastructure or an essential service that cannot realistically be allowed to fail.
If the answer is yes, there should be a presumption that the Government retains substantial in-house capability.
It could still use private contractors where specialist skills are needed.
But the Government should remain capable of running the system itself.
That is a very different model from simply handing the entire responsibility to a contractor.
Government needs a "right to fail"
Perhaps the most important principle should be this:
No critical government service should ever be dependent upon a single private supplier without a credible alternative.
If Palantir disappeared tomorrow, could the NHS continue operating its data platform?
If Capita walked away from a major contract, could government take the service over?
If a technology supplier went bankrupt, could the Government access its data and systems?
If a contractor suffered a major cyberattack, could the service continue?
If the answer to those questions is no, the Government has created a strategic vulnerability.
And that vulnerability may not show up on the original procurement spreadsheet.
There is another issue: profit
This is where your earlier point about Scottish Water becomes relevant.
There is nothing wrong with a company making a profit.
But if the Government is paying a private company to deliver an essential public service, it should be able to demonstrate that the profit represents genuine efficiency rather than simply a transfer of public money to shareholders.
And if a contractor repeatedly fails, the public should not be expected to finance the cost of putting things right while the contractor continues to win new government business.
That is particularly difficult to justify in cases where the Government has already accumulated evidence of poor performance.
Britain needs to stop thinking in terms of "public versus private"
I think the debate would be much more productive if we stopped treating this as a simple ideological battle.
The real question is:
What is the most effective ownership and management structure for each particular service?
Sometimes that will be private.
Sometimes it will be public.
Sometimes it may be a hybrid.
But for critical services, I would argue that government must always retain the expertise and infrastructure necessary to take control if the private provider fails.
That changes the balance of power.
If a contractor knows that the Government can take the service back, it has a much stronger incentive to perform.
If everyone knows that government has no alternative, the contractor has considerably more leverage.
The Government Shared Services programme offers another warning
This isn't just about famous companies.
The Public Accounts Committee warned in July 2026 that the Government's own multibillion-pound programme to consolidate back-office functions was already suffering from delays and weak oversight. The programme had been expected to deliver £4.3 billion of benefits but had been delayed by five months, with the PAC warning that the Government needed to reconsider its approach to prevent a costly failure.
That illustrates another important point.
Bringing something in-house doesn't automatically make it successful either.
Government can design bad systems.
Government can waste money.
Government projects can go over budget.
The answer therefore isn't to assume that public employees are always better than private employees.
The answer is to create strong institutional capability, proper accountability and long-term responsibility, regardless of who actually carries out the work.
Perhaps Britain outsourced too much of its memory
This may be the biggest lesson.
For decades, governments have concentrated on reducing headcounts and buying services from companies that supposedly possess expertise that government no longer needs to maintain.
But expertise is not a commodity that can always be bought back when required.
Once a government department loses 20 years of accumulated knowledge about a complex system, it cannot necessarily recreate that knowledge by issuing another contract.
And that is why the Capita pension problem is more important than simply another outsourcing failure.
The Government is now having to ask whether it can run a service that it previously decided it did not need to run itself.
The Post Office scandal demonstrated what can happen when government cannot adequately challenge a technology supplier.
The Palantir debate demonstrates what can happen when a critical public service becomes heavily dependent on a foreign technology provider.
These are different cases, but they point towards the same underlying issue.
The state needs to know how its own machinery works.
So should Britain bring more services back into public control?
Yes — but selectively and intelligently.
I would not advocate bringing every outsourced service back into government.
But I think there should be a strong presumption that critical public services, sensitive data systems and functions requiring long-term institutional knowledge should retain a substantial in-house capability.
Where a private contractor repeatedly fails, the Government should seriously consider terminating the contract rather than continually renegotiating it.
And where the Government has to rescue the service after failure, the cost of that rescue should be included when calculating whether outsourcing actually represented value for money.
That last point is crucial.
A £100 million contract that eventually costs another £100 million to repair after failure was not a £100 million contract.
It was a £200 million decision.
And if the Government has to spend years rebuilding the expertise it previously discarded, the real cost could be considerably higher.
The British state does not necessarily need to become bigger.
It needs to become more capable.
That is a very different proposition.
The lesson from Capita, Fujitsu, Palantir and other large government technology and service contracts should not be that private companies are inherently bad.
It should be that government should never outsource its ability to understand, manage and, if necessary, take back control of something upon which the public depends.
After all, when the contractor fails, the contractor may lose a contract — but the taxpayer is left with the consequences.
And perhaps that is the question that should have been asked before many of these contracts were signed:
"If this goes badly wrong, can we take it back — and do we still have the people who know how to run it?"
If the answer is no, perhaps the Government has not really outsourced a service at all.
It has outsourced responsibility while retaining the risk.