Ten years ago, transferring money in the UK was a process fraught with fiddly gadgets and fat-finger danger. Banks weren’t able to verify your identity across apps, so they gave you proprietary hardware in the form of a physical card reader. Naturally, multiple bank accounts meant multiple card readers. You also had to manually enter the recipient’s name, account number and sort code to send money. But there was no verification of that information, so if you accidentally got it wrong, even by one digit, the money shot off into the digital void, potentially ending up in a stranger’s account. Getting it back was a lot of paperwork. 

Today, banks verify your identity instantly using your phone’s built-in biometrics. Account details of payees are pre-verified, so you know money is going to the right place. If you have accounts with different banks, you can view your balances within a single banking app. And if you want to switch banks, you can port your entire financial identity, from direct debits to years of transaction history, to a new provider in just a few taps. It’s easier, faster, and more secure, and all possible thanks to Open Banking.

Open Banking demonstrates the power of digital public infrastructure (DPI), the set of digital ‘rails’ on which both public services and private innovation run. When you buy a new toaster, you don’t expect to rewire your kitchen or change your wall socket; the voltage is standard, the plug is uniform, and power from the grid is a given. DPI provides the same universality for the digital age, functioning as a shared operating system rather than a standalone app. It provides predictable infrastructure on which digital goods and services can be built.

This foundational quality is what makes DPI distinct from other policy interventions. It doesn't just improve one service or sector, it underpins all of them. Planning reform, energy transition and healthcare innovation all depend on data moving freely between systems. You cannot build a smart grid, digitise the planning system or unlock NHS data for research without first fixing the underlying rails.

DPI unlocks growth through three core mechanisms: eliminating the friction tax, enabling public and private innovation and treating data as productive capital. The UK’s current ‘tenant state’ model, characterised by vendor lock-in and a sovereignty gap, is actively stalling these drivers of growth. Transitioning from a digital tenant to a sovereign owner of these foundational rails offers a multi-billion-pound prize, evidenced by home-grown and international success stories, from the UK’s Open Banking to Estonia’s X-Road and the India Stack.

The most widely cited DPI success story is India's 'India Stack', a suite of open, interoperable systems covering digital identity (Aadhaar), payments (UPI), and data exchange (account aggregator) that any business or citizen can build on. Building this infrastructure allowed the Indian state to improve government services while unlocking private innovation at scale. In practice, it means an ordinary citizen can scan their fingerprint to verify their identity and activate a SIM card, or buy a cup of tea using a QR code that draws from whichever bank account they choose. 

India Stack’s three pillars - identity, payments and data exchange - are typically considered the foundation of any DPI initiative. Some definitions are broader and also include trust infrastructure (such as signatures) and discovery and fulfillment (such as open APIs for tax filing or business registration). 

What makes DPI distinct from digital infrastructure that happens to be publicly used is sovereignty, openness and interoperability. These three properties are what allow DPI to function as genuine public infrastructure, rather than a public-use private service.

The success of Open Banking is the exception rather than the rule for DPI in the UK. While India has spent a decade demonstrating what a sovereign identity, payment and data exchange stack can do for population-wide growth, the UK’s public sector data infrastructure remains trapped in a ‘tenant state’ model. Systems are locked into proprietary vendor contracts, meaning the government effectively pays rent to private companies to access its own data. Accessing, using, and linking data is a slow, manual, expensive process.

AI has fundamentally changed the value proposition for fixing this. Powerful AI won't work without good data infrastructure. Giving health researchers the most advanced AI tools does not solve the issue of all 6,200 GP practices in England acting as individual gatekeepers to patient records. Similarly, equipping local authorities with premium LLMs does not address the fact that the single best record of a child’s early years development is still a physical book.

Digital infrastructure has long been treated as a back office maintenance issue, deprioritised and underfunded. It is now the key to a generational reboot of public services and economic growth. With the most advanced tech sector in Europe, the UK economy stands to gain hundreds of billions of pounds over the next decade by capitalising on AI, with potentially tens of billions of pounds per year in public sector savings.

These gains are not guaranteed. They are contingent on good underlying data - machine-readable, interoperable, portable. Despite having the third largest AI market in the world, much of the UK’s data assets are trapped in 20th century silos. Critically, these silos are often owned and controlled by foreign tech companies, creating a sovereignty gap where the state lacks the legal and technical authority to use its own data. If we don’t modernise our protocols, standards and governance, existing inefficiencies will simply be automated and dependency on overseas providers deepened.

Without addressing the underlying data, AI becomes an expensive layer of paint on a crumbling wall. The private sector understands and invests in designing and exploiting data systems. The public sector must stop leaving this value on the table. DPI offers the opportunity to achieve this.

There are three main mechanisms through which DPI creates economic value. In the UK, however, these mechanisms are currently stalled.

Mechanism one: eliminating the ‘friction tax’

When data is held in incompatible, proprietary or inaccessible systems, every transaction that requires that data becomes slower, more expensive and more error-prone. It is a structural drag on economic output, acting as an invisible tax levied on every interaction between citizens, businesses and the state.

In countries where DPI has been successfully deployed, this tax is eliminated at source. When identity, payments and data exchange run on shared open rails, the cost of each transaction falls dramatically. For example, the introduction of Aadhaar and UPI in India cut the cost of opening a bank account from $20 to $0.27. Removing the need for manual verification and physical paperwork saved money, while also bringing millions of citizens into the formal economy.

The UK's fragmented data landscape, by contrast, imposes this friction tax extensively. Citizens currently spend an estimated week and a half each year dealing with government bureaucracy. Much of it is caused by systems that cannot talk to each other: moving home requires contacting ten different government departments; managing a long-term condition or disability means navigating more than forty services across nine organisations; there are forty-four different ways to prove who you are.

In the energy sector, 70% of meters in UK homes are smart meters, yet just 2.8% of households are on smart time-of-use (ToU) tariffs, which reward shifting usage to off-peak periods and can save households £200 a year. Research suggests households are more likely to switch to these cost-saving tariffs if they can ‘set and forget’ - automatically shifting usage to cheaper periods without manual effort. But this requires smart meter data, appliance systems and suppliers' pricing engines to communicate in real time, something proprietary silos currently prevent. 

In healthcare, the consequences of friction tax are grave still. In England, an estimated 1.8 million undetected drug errors occur during transitions of care, when patients move from one healthcare setting to another (for example from an acute hospital ward to community care). This is estimated to cost £17.8 million and contributes to avoidable harms for more than 30,000 patients. A single shared digital prescribing record could improve interoperability across different care settings, helping to reduce information gaps and support safer, more coordinated transitions.

Unlocking this mechanism in the UK would be transformative. The impact is evident just from these examples - giving citizens hundreds of hours of time back per year and cutting national energy bills by billions - effectively lowering the cost of living through better digital plumbing. 

These examples demonstrate the immediate savings from removing friction. The far larger prize is what becomes possible once the underlying rail exists. A smart energy system built on truly interoperable data does more than save households £200 a year on their bills. It becomes the foundation for new energy services and innovations not yet imagined. This is the pattern that follows successful public infrastructure investment: the returns that justify the initial cost are dwarfed by the innovation that follows.

Mechanism two: enabling public and private innovation

When the state builds and owns the underlying infrastructure, it creates a platform on which government departments and private actors can build innovative services, removing the need for new projects to start with the plumbing. In the public sector, this breaks down departmental silos, allowing for rapid, cross-cutting services. India achieved 80% financial inclusion 40 years faster than it would have done without DPI.

For the private sector, this logic produced the internet, GPS and, more recently, Open Banking. None of these generated their economic value by being used directly; they generated it by enabling a superstructure of private innovation that would not otherwise have existed.

Open Banking has added an estimated £4 billion to the economy and enabled the creation of almost 5,000 high-skilled digital jobs. India's UPI now handles over 20 billion transactions per month and has acted as a massive catalyst for private credit. From 2023 to 2025, Indian districts with high adoption saw a 10x higher growth rate in consumer loans, as payment data allowed private lenders to verify the creditworthiness of millions of new borrowers. Brazil's Pix, a state-owned instant payment system built for $4 million, had delivered estimated savings of $5.7 billion by 2021 and is expected to contribute 2% to GDP growth by 2026. These are prime examples of private innovation on sovereign public rails. 

The condition for this mechanism to work is that the rails must be genuinely open and interoperable. If the underlying infrastructure is proprietary, public departments and private actors cannot build freely on top of it. They must instead negotiate access, pay licence fees and accept terms set by whoever controls the infrastructure. 

The UK's life sciences sector illustrates what is lost when this mechanism fails. Industry-led clinical trials fell by 41% between 2017 and 2021. Fragmented data means commercial sponsors lack a single view of which NHS sites have capacity to run a trial or where relevant patient populations are located, so they select sites 'in the dark'. The cost to the economy has been £2.6 billion per year since 2017. Conversely, Denmark's DANFLU-2 trial used linked national health registries to enrol 332,438 participants at a speed and scale currently unachievable in the UK.

Mechanism three: data as productive capital

When data is interoperable and portable, it can function as a productive asset, like financial collateral, research infrastructure or evidence for targeted interventions. 

In India, a street vendor with no house or car can use their digital sales record to prove they are a reliable borrower, securing a micro-loan on their phone in seconds. The UPI payment record has become a financial asset. The data was always there, but the infrastructure made it legible, portable and trustworthy.

In the UK, there are localised examples of how DPI can benefit policymaking. The Born in Bradford programme links health, education, social care and environmental data for 800,000 individuals across forty years. By linking data, researchers have found novel evidence that has led to targeted interventions. For example, they have demonstrated the harm of traffic-related air pollution on childhood brain development, triggering interventions to improve air quality.

Flooding costs the UK £2.2 billion per year, and more than one in six properties in England are at risk. By linking business and flood-risk data using unique property reference numbers (UPRNs), the Bank of England can precisely identify firms exposed to flooding and assess financial vulnerability. This helps policymakers target resilience measures that could mitigate severe impacts, including 35% drops in assets, employment and turnover among affected firms.

AI has increased the urgency of addressing this mechanism. Machine learning systems require large volumes of clean, interoperable, machine-readable data. The UK's fragmented landscape means that even where data exists in abundance, it cannot be fed into AI systems at the scale needed to generate the productivity gains the government is projecting.

Understanding why these three mechanisms are failing in the UK requires understanding how the UK's data landscape came to be the way it is.

The UK’s fragmented data ecosystem is a byproduct of the ‘tenant state’ model that has emerged following decades of government IT procurement practices. The 20th-century mantra of 'nobody got fired for buying IBM' has evolved into a modern dependency on a handful of dominant data platforms. This has traded short-term stability for long-term vendor lock-in and a hollowed-out internal technical capability. The result is the government pays to store its data in proprietary vendor systems it cannot easily access or move.

The Department for Science, Innovation and Technology’s 2025 State of Digital Government Review lays out just how outdated the current infrastructure is. An estimated 28% of central government departments rely on legacy systems; only 27% of the Review’s 131 public sector survey respondents believed their current data infrastructure enables a comprehensive view of operations or transactions; and 70% say their data landscape is not well co-ordinated, interoperable, or enables a unified source of truth.

When data is held in proprietary systems, the government is ‘locked in’, forced to pay vendors high API fees just to access its own information. Linking data across systems is a further challenge as it is held in proprietary formats and vendors lack incentives to increase interoperability. The result is a bureaucracy tax for citizens and a data poverty trap for the state.

The Horizon post office scandal serves as a stark reminder of what happens when the state cedes control of its data infrastructure to proprietary, opaque vendor systems. It loses the ability to audit the truth, in this case with severe human consequences as more than 900 sub-postmasters were wrongly convicted. Similarly, The Police National Computer (PNC) has been running since 1974. After a decade of attempts to replace it, costing over £1 billion, the government remains tethered to its 1970s architecture built on proprietary databases. The data is stored in a non-standard format that is prohibitively expensive to migrate, which has real-world consequences. In 2021, 150,000 records were accidentally deleted due to a single coding error, completely avoidable in modern systems.

The crisis of the tenant state is now entering a more dangerous phase. Modern procurement, epitomised by the £330 million Palantir Federated Data Platform in the NHS, risks replacing 20th-century vendor dependency with 21st-century digital feudalism. While the UK government may own the data, foreign companies own the logic (the algorithms, the ontology and interface). That means the government cannot see or use the data without paying foreign companies licensing fees, API costs or consulting hours. These costs are significant: in 2023, 55% (£14.5 billion) of public digital and data budget was spent on contractors, managed services providers, and IT consultants, compared with 20% (£5 billion) for permanent staff. Where vendors previously owned the filing cabinets, now they also own the ability to read the files.

The reliance on foreign vendors also cedes jurisdictional control. For example, under the US CLOUD Act, critical British infrastructure that has been outsourced to US tech firms remains subject to the reach of US law. The UK’s public services cannot be sovereign under these circumstances, where the state's data is governed by a foreign provider’s proprietary terms, leaving critical infrastructure exposed to ‘kill switches’ triggered by shifting geopolitical winds.

There is growing consensus around the economic case for data infrastructure upgrades in the UK, as demonstrated by the government’s recently published Smart Data 2035 strategy. Backed by the Data (Use and Access) Act 2025, which grants the state statutory powers to mandate data sharing, the strategy sets out an ambitious roadmap for twenty ‘smart data schemes’ across sectors, from agrifood to digital markets. 

Government modelling suggests that smart data schemes in just four sectors (housing, international trade, retail and energy) could generate £71.2 billion in net social value by 2043, with annual GDP contributions reaching £9.6 billion.

These figures are significant, but represent only one stream of the available prize. Smart data schemes focus on making privately-held data more interoperable for consumers, by mandating the sharing of customer data with authorised third parties at the customer’s request. Open Banking demonstrates what a regulatory mandate for privately-held data like this can achieve. But it also emphasises why DPI must advance in parallel. It was the underlying sovereign infrastructure of shared identity verification, open payment rails and standardised APIs that enabled Open Banking to deliver value at scale. As smart data schemes extend across other sectors, the same logic applies: the value of opening up private data is contingent on having sovereign public rails to carry it.

The wider benefit from building those rails is substantial. A landmark study by the Open Data Institute estimates that adoption of open data frameworks across core public data assets would provide 0.5% of GDP more economic value compared to restricted or paid models, or around £15 billion per year. Again, this captures only the benefit of making public data free to use, not the full systemic value of owning the rails on which it runs. 

A full DPI model goes further; it establishes the sovereign protocols for identity and exchange that prevent the state from becoming trapped in a cycle of vendor lock-in and ‘API rent’, enabling AI at scale and creating the foundation on which public services and private innovation can compound. These estimates demonstrate that even partial fixes, like making it easier to move data, is worth tens of billions of pounds. To capture the full prize, the UK must move beyond mandated sharing and start owning the foundations of its digital economy.

The argument for DPI is an argument for national growth and resilience, and it is one the government can act on now. Fixing our digital plumbing provides a rare political trifecta: immediate fiscal savings for the public sector, a massive competitive advantage for key industries, and a tangible improvement in the daily lives of millions. Unlike roads or railways, digital infrastructure can be built at speed and relatively low cost; Brazil built Pix for $4 million. The UK does not face a decade-long construction programme, but a choice about whether to act.

The Smart Data 2035 Strategy and Data (Use and Access) Act represent genuine momentum. But smart data alone is not sufficient. The value of that data is contingent on having sovereign public rails to carry it. As long as the state remains a tenant in someone else's system, paying rent to access its own data, the full prize remains out of reach. This problem has been years in the making, but AI has significantly magnified the potential reward for solving it.

Building a sovereign DPI requires a series of structural changes spanning how the state procures technology, funds infrastructure, rebuilds its own technical capability, governs public data and mandates open standards. Alongside these foundations, a set of near-term opportunities demonstrates that meaningful returns are achievable using data and standards that already exist. The question is not whether we can afford to build this. It is whether we can afford not to.