Building infrastructure for the AI age

Bangkok Post

6 days ago

The London Underground, the world's oldest subway system, opened in 1863. Around the same time, London's modern sewage system was designed by civil engineer Joseph Bazalgette in response to the Great Stink of 1858, which brought parliament to a standstill. Planning far ahead, Bazalgette built the system to last 150 years. Only now, with the Thames Tideway project, is it being significantly expanded.
Walk through any major city in the UK -- from London to Glasgow, Belfast to Bristol -- and you will find that much of the infrastructure from the late Victorian era is still in use. That is partly because the Victorians built early and planned decades ahead. Like in many other countries, the UK's infrastructure is akin to a palimpsest, with new layers constructed over the old.
A similar dynamic is playing out today, as new technologies become integral to daily life and economic growth. Broadband, mobile networks, and data centres, which are now as vital as roads and power grids, are prime examples.
At the same time, the concept of social infrastructure is gaining traction among policymakers around the world. The UK's new ten-year infrastructure strategy, for example, focuses on "opportunities for collaboration, productivity and efficiency gains, and the wider benefits of strategic and spatial planning" across the health, education, and justice systems. As the plan notes, this is the first time the British government has included social infrastructure in its national strategy.
But even that view is too narrow. Consider the characteristics of the systems that keep our economies running: long lifespans, high fixed costs, low marginal costs, and broad accessibility. Importantly, their value is derived not from the physical assets themselves, but from the economic activities they enable.
Our definition of infrastructure should be expanded to reflect the demands of the digital age. While governments often view AI infrastructure in terms of data centres and the energy and water they consume, a truly comprehensive view must also include intangible assets such as software and data.
To be sure, this might seem like a conceptual leap. But the Covid-19 lockdowns of 2020-21 revealed just how essential digital platforms have become. Videoconferencing tools like Zoom, for example, are so critical to work, education, and public services that it is difficult to imagine life without them.
As a result, a growing number of countries are developing what is often called digital public infrastructure, though in practice it typically involves a mix of public and private services. In 2020, Brazil's central bank launched the Pix real-time payment system, which has largely replaced cash transactions. Similarly, India's Aadhaar biometric identification system now serves as a platform for both public services and digital payments.
Such digital systems have been adopted more quickly in countries with fewer entrenched legacy services. By contrast, countries like the US and the UK have long-established payment systems dominated by private providers, such as credit card companies, which can impede the adoption of public alternatives.
Given their growing economic importance, governments must start thinking strategically about software and data -- the digital equivalents of roads and power grids.
This is especially true of intangible assets. Cash-strapped governments tend to underinvest in resources like data and software, while private investors often view them as too risky. But the lack of a robust and well-maintained digital foundation can hinder economic growth.
A second reason to invest in digital infrastructure is national sovereignty. Over the past few years, policymakers around the world have become increasingly concerned about the national security risks posed by US firms' dominance in cloud computing.
The lesson for governments is that they must step back from immediate flashpoints and take a longer view. Infrastructure provides a useful lens for thinking strategically about what investments are needed, who should make them, and how they should be governed to sustain economic growth.
Equally important is a foundation of high-quality data and interoperable software that facilitates user authentication, improves access to cloud services, and fuels the creation of new digital businesses.
Victorian planners' foresight continues to benefit us more than a century later. We should approach today's infrastructure challenges with the same mindset. ©2025 Project Syndicate
Diane Coyle, Professor of Public Policy at the University of Cambridge, is the author of 'Cogs and Monsters: What Economics Is, and What It Should Be' (Princeton University Press, 2021) and 'The Measure of Progress: Counting What Really Matters' (Princeton University Press, 2025).