The officials who signed off on the Channel Tunnel in 1987 were not fools. They followed the rules: rigorous analysis, approved discount rates, stress-tested scenarios. Within a decade, the project was renegotiating its debt. By 2006, it had gone through two rounds of debt restructuring. Why hadn’t the appraisal framework spotted the risks of this happening? Because the framework itself was the problem.
The same pattern runs through more recent major infrastructure decisions. HS2 committed tens of billions before the northern legs were cancelled. The 2022 European LNG build-out added urgent capacity after the energy shock – and now runs at around 52% utilisation as demand falls short.
The pattern extends further: £1.15bn committed across 25 interdependent government shared services programmes, with no mechanism to redirect funds as priorities shifted. One architecture links them all: a one-time bet on a long-range forecast that cannot be revised once the commitment is signed.
How the current system works – and why it fails
A major infrastructure decision works roughly like this. Analysts produce a demand-and-cost forecast stretching decades ahead. That forecast feeds into a business case. Once the business case is approved, the commitment is locked in – often for decades. The analysis that justified the decision does not travel with it: what gets funded is the project, not the ongoing question of whether it still makes sense.
Reversing that commitment later, even when circumstances change beyond recognition, is extraordinarily difficult. The National Audit Office found in March 2025 that of the government's 108 most significant major projects, representing £432bn in spending, only nine had robust evaluation arrangements. For 64% of that spending, there were none. The forecast does not inform the decision so much as replace it: once the number is accepted, the governance architecture offers almost no route back.
The OBR's March 2026 outlook made the price of that rigidity unusually plain. Debt servicing has risen from £39bn in 2019–20 to £106bn last year – not only because of rising health and welfare costs or broader fiscal pressures, but because pre-committed obligations proved impossible to unwind when the underlying assumptions changed. The pattern is not occasional. Borrowing has remained around 5% of GDP for four consecutive years, persistently higher than forecast.
Why the next Spending Review matters – and why preparation starts now
SR25 has already moved in this direction: Integrated Settlements for seven Combined Authorities replaced ring-fenced grants with flexible multi-year allocations, and GDS/HMT pilots launched in September 2025 replaced one-off project approval with continuous portfolio allocation across HMRC, NHS England and DSIT. SR27 is the next opportunity to extend the same logic to national infrastructure. But infrastructure decisions move slowly: the projects entering development now, appraised under today's framework, will arrive at SR27 ready for commitment. By then, the architecture will already have done its work. The window to change it is now.
What would change?
Annual spending envelopes. Currently, each project is appraised in isolation against a multi-decade forecast. Instead, the government would set an overall infrastructure spending limit and allocate across competing priorities each year. The question shifts from "Does this forecast justify a thirty-year commitment?" to "Does this project deserve priority this year, given everything else in the portfolio?" By the time PFI was abolished in 2018, over 700 contracts signed since the 1990s had locked in future commitments of roughly £199bn – largely invisible to routine scrutiny. Annual envelopes make the same spending decisions explicit, contestable and revisable.
Staged commitments with genuine exit points. Currently, approval is largely binary: funded or not. Instead, initiatives would move through stages, each delivering something useful, with a real decision point before the next tranche is released. Projects that cannot be broken into meaningful phases would be flagged as one-way bets – and scrutinised accordingly. HS2's trajectory – tens of billions committed before cost overruns became politically undeniable – is the model to avoid.
Equal treatment of build and maintain. The current accounting framework favours new construction over maintaining existing assets. A new bridge counts as an investment. Sensors and maintenance that double the life of the existing bridge count as running costs. The bias is not just wasteful – it locks forecast errors into concrete, literally and irreversibly. Italian energy regulators have already removed this bias, setting spending targets that treat capital and operational solutions on equal terms.
What this means in practice
Take any business case currently on a programme board in your directorate. Ask how it would look structured as an annual allocation, revisable next year, a staged commitment with defined exit points, or indifferent between a new build and an operational alternative. If all three collapse into the same proposition – a one-time bet on a thirty-year forecast – the business case is not asking the right question. Governments that cannot redirect resources when the world changes are not governing. They are administering the consequences of past assumptions – and paying £106bn a year for the privilege.
Spending Review 2027 is less than two years away. The project pipelines being assembled now will determine what it funds – and how.
Vsevolod Shabad is a principal enterprise architect and independent researcher focused on public investment governance. Ruslan Yusupov is an enterprise agility practitioner specialising in lean portfolio management. Kirill Ivanov is a strategy and innovation executive focused on capital allocation in asset-intensive organisations.