The Continuum+: Powering the AI era

Powering the AI era

When J. Presper Eckert and John Mauchly created the first data centre in the 1940s, they could hardly have imagined it would become the engine of a future shaped by artificial intelligence (AI). But today, data centres are powering growth on a scale that’s straining grids, land and planning systems like never before.

No longer just a technology story, the sector now sits at the intersection of energy, infrastructure and finance, where decisions in one area ripple across the others. With demand continuing to soar, can our infrastructure scale fast enough to keep pace while sustaining the people and resources that make it possible?

AI demand is growing, but the economics are unresolved

Electricity demand from data centres is set to more than double by 2030, reaching roughly 945 terawatt hours a year – about the same as Japan’s annual consumption today. Much of this growth will be driven by AI-optimised servers, creating new pressure points across power markets, cities and global supply chains.

This has added a new, voracious layer to what was already a surging cloud driven appetite for data centre capacity. In parts of the US and Europe, they consume over 20% of local power, forcing hard choices for businesses, governments and regulators. Forecasts suggest that meeting AI related demand alone will require $5.2 trillion in investment by 2030, on top of the rest of the industry’s needs.

Yet for all this capital, the underlying economics remain unsettled. “The big question is: is AI actually going to make money?” asks Barry Gross. “Building an AI-ready data centre costs tens of millions of pounds per megawatt, and that compute power only lasts a few years. Can revenue from end users cover that? If not, then it will all come crashing down.”

The answer is unclear. On the cost side, the physics of AI infrastructure – high density racks, power hungry graphics processing units (GPUs) and advanced cooling – drive up capital and operating expenditures. On the revenue side, earnings remain closely tied to broader platform use rather than standalone AI applications. Direct monetisation of AI is still emerging, with companies experimenting with subscription or usage based pricing.

Until that revenue model is clearly proven, the industry faces an unresolved tension: huge growth in demand, but economic models that have yet to catch up with the scale of investment needed to feed it. This is already reshaping the way capital markets respond. With individual projects costing billions, lenders are building dedicated teams to focus on digital infrastructure.

“What’s changing is the way risk is allocated,” explains Céline Larmet. “Lenders have been increasingly syndicating facilities and bringing in private credit and institutional investors at an earlier stage, because no single balance sheet wants to carry that exposure alone. Data centres are also large-scale, distinctive assets, sitting at the crossroads of real estate and infrastructure when it comes to financing risk analysis.”

The new geography of data and AI infrastructure

It’s not just the economics of data centres that are in flux; where and how data is processed is becoming just as critical as how it is monetised. Governments are increasingly insisting that data be stored and processed locally, transforming what was once a purely technical decision into a matter of national policy and strategic importance.

“Sovereignty has quietly become one of the most important ideas shaping modern technology policy. A few years ago, everyone was talking about data sovereignty and then sovereign cloud. Today, it’s about AI sovereignty,” observes Raza Rizvi. “Too much reliance on certain countries for critical technologies and vendors is raising concerns about digital dependency.”

Across Asia, Europe and the Americas, a patchwork of data localisation and sovereignty laws is taking shape. China’s cybersecurity and data security laws require certain data types to be stored and processed onshore; Russia mandates the local storage of its citizens’ personal data; and even in Europe, where GDPR does not explicitly demand localisation, strict transfer rules effectively tether sensitive data to the EU. Across Africa, Latin America and parts of Asia, similar frameworks are emerging or being strengthened.

These regulatory and policy factors are especially complex in China. “China’s approach to data and AI governance reflects both national security priorities and industrial strategy,” observes Jingyuan Shi. “Companies must navigate stringent localisation requirements, approvals for foreign investment, and evolving cybersecurity standards, making digital infrastructure investment in China a highly strategic and sensitive undertaking.”

Minesh Tanna adds a broader perspective: “The prospect of international harmonisation on data and AI laws seems unlikely. Each region is approaching AI and digital infrastructure from very different angles, which has a huge impact on how opportunities and risks are perceived.”

These rules are increasingly shaping the strategic landscape for digital infrastructure. In seeking greater control over how data, and by extension AI, is governed, policymakers are beginning to view digital infrastructure more as a sovereign asset rather than a global utility. For example, Germany has moved to build a sovereign cloud for its public sector that keeps data and operations physically and legally within the country’s borders to meet local sovereignty requirements.

Yet localisation can be double edged. Although it promises legal control and reduced dependence on foreign jurisdictions, it can also raise barriers to entry for smaller firms and complicate cross border operations. Fragmented regulatory approaches are therefore forcing businesses to think more carefully about where they host their data and where they carry out processing activities.

“It no longer makes sense for businesses to put all their eggs in one basket,” says Andrew Joint. “We’ve been moving towards a model where everything sits in a single cloud, with one provider doing everything, and that’s not ideal for competition or resilience. If that provider is tied to a particular government or market, the risks increase. As a result, organisations may fragment their service provision, even if it introduces user-experience detriment or operational friction.”

Stretching the limits of power and people

Beyond these legal and regulatory hurdles, the practical limits of running large-scale digital infrastructure are becoming more complex. And increasingly, they’re defined by a lack of access to power: grids rarely have the capacity to meet demand in the locations where hyperscale facilities are needed.

“Power supply restrictions are increasingly driving where it’s feasible to build data centres at all,” says Jaap Tempelman. “In the Netherlands, permits have become much harder to obtain, with strict requirements on energy efficiency, location and sustainability. That’s creating a real tension between the need to expand digital infrastructure and the limits of the grid.”

Operators are experimenting with creative solutions. Some are tapping unused office power, linking distributed assets at a small scale across cities. Others are going even further, using cutting-edge technology to optimise operational efficiency before a single server is deployed.

“We’re advising a client that’s creating a digital twin of an entire country,” says Raza. “And within that, they’re developing sophisticated virtual representations of data centres and other digital infrastructure. That’s allowing them to capture vast amounts of data linked to operational efficiency and achieve ambitious optimisation goals at pace and at scale.”

Space-based infrastructure has made headlines recently, but it remains a niche, high-risk option. Satellite links are bandwidth-limited, expensive and vulnerable to launch failures, making them no substitute for terrestrial data centres.

“There’s no solution in space that can match the capacity of fibre,” explains Waltter Kulvik. “The likelihood of a satellite failing during launch or otherwise malfunctioning is also higher than most people realise, which affects time to deployment.”

Talent, meanwhile, is another critical constraint. The industry’s ageing workforce – the average data centre professional is 54 years old – struggles to keep pace with rapid expansion. Expertise in power engineering, construction, operations and finance is in short supply, forcing operators to invest heavily in training and early engagement.

“When it comes to scaling data centres, the biggest issue isn’t regulation,” says Waltter. “It’s having the capacity and the right people to manage it.”

ESG: An easy target, a complex reality

The soaring power demands of data centres are putting ESG squarely in the spotlight. In regions where grids are strained or electricity comes at a premium, operators face a high-stakes balancing act: grow fast, but do so efficiently. Every infrastructure decision, from where to build to how to operate, carries direct implications for sustainability.

Global business drives much of the load through cloud computing, AI training and enterprise applications. But consumer behaviour is equally impactful: streaming video, online gaming, social media use and video conferencing all contribute significantly to the energy footprint. “We live in a demand-driven world,” says Barry. “And we have to accept that using technology comes with a cost.”

Changing consumer behaviour, however, is far from straightforward. Although many people express concern about sustainability, research shows this doesn’t always translate into day-to-day decisions. Their choices are still dominated by convenience, habit or cost.

“Should individuals be responsible for making a product as efficient as possible?” asks Andrew. “In reality, environmental concerns only translate into action where they’re enforced by governments. That environmental obligation is likely to fall on those who are deemed able to afford it: energy providers and data centre operators.”

To address these pressures, operators are experimenting with a mix of nuclear, private generation and renewables. In Singapore, where much of the power is imported, data centres are backing floating renewable energy initiatives to cut carbon while maintaining reliability. Elsewhere, in Sweden, EcoDataCenter’s Falun campus runs entirely on local hydropower and regional wind, showing how facilities can operate at scale using 100% renewable energy.

But these solutions are still early-stage. Storage, compute and cooling needs continue to grow faster than clean supply can keep up with, and integrating new energy sources at scale remains complex and costly. There’s a long way to go before data centres can fully decouple growth from environmental impact.

To add to these technical challenges, regulators worldwide are introducing new energy and sustainability rules for data centres. Singapore has launched a national standard requiring IT equipment to meet minimum efficiency targets. In the US, states like California are moving toward disclosure and efficiency obligations for operators. And in Europe, the revised Energy Efficiency Directive raises the EU’s overall energy efficiency target, requiring member states to collectively deliver an additional 11.7% reduction in energy consumption by 2030, while introducing “energy efficiency first” as a fundamental principle of EU energy policy.

“There’s a balancing act at a regulatory level,” says Jaap. “On one hand, the EU wants to incentivise investment in AI and digital infrastructure. On the other, it’s introducing stricter sustainability and energy efficiency requirements. Data centre operators have to navigate both at the same time.”

Financing the next wave of digital infrastructure

The scale and pace of digital infrastructure growth mean tough decisions lie ahead. And none are more pressing than funding the next generation of AI ready data centres. Compute-intensive racks, tariffs and supply-chain constraints are driving capital costs ever higher, making investment strategy just as critical as powering them efficiently.

“There’s a huge wall of capital needed to develop data centres, and there’s only so much lenders can finance on their balance sheets,” warns Kirsty Barnes. “As lenders get more excited about the sector, there needs to be a convergence of different lending platforms because banks alone can’t fund it all.”

Data centre financing in Asia is shifting rapidly from traditional bank loans to specialised, high-volume funding from private equity, institutional investors, and green loans to support a projected US$100 billion in new construction over the next five years, driven by AI demand. Key hubs include the Johor-Singapore-Batam corridor, where AI-ready infrastructure is increasingly funded via structured finance, asset-backed securities, and sustainability-linked loans.

Against this backdrop, fundamentals matter just as much as financial structure.

Calvin Tan explains: “In Asia, the success of a data centre depends on far more than its physical footprint. While energy efficiency matters, access to reliable power and, crucially, water is what sets projects apart. In tropical climates, water for cooling high performance AI systems has become the key bottleneck – and those that solve for it are attracting the strongest investor interest.”

Recent industry forecasts highlight just how intense the investment challenge has become. Global data centre capital expenditure could reach about $1.2 trillion annually by 2029, driven in large part by AI optimised infrastructure and specialised hardware.

“In Europe, we are starting to see a shift towards treating data centres more like infrastructure assets,” observes Kirsty. “There’s a greater focus on revenue streams, rather than just the underlying real estate.”

Recently, a major European data centre operator, DATA4, raised €3.3 billion to refinance its existing facilities and support further expansion. Simmons & Simmons advised DATA4 on the transaction, one of the largest financings in the European data centre sector to date.

As capital requirements grow, lenders are exploring new financing approaches, including securitisation, to broaden the investor base and free up bank balance sheets. However, the legal and operational complexity of data centres means this requires careful structuring.

“The challenge lies in legally ring-fencing the assets and protecting income streams in the event of a potential insolvency at the data centre operator,” Jennifer Aubry explains. “Unlike, for example, residential mortgages, which largely take care of themselves through direct debits, a data centre requires constant management of power, security, service continuity and staff. This makes contractual ring-fencing more complicated than in other, more granular securitisation asset classes.”

At the same time, private equity and other investors are changing the way deals are done. They’re increasingly willing to pay high multiples for companies supporting digital infrastructure, betting on long-term growth even when profits remain uncertain. Competition is fierce, and the highest bidder doesn’t always win.

“It's not necessarily just about being the highest bidder for an asset now,” says Geraint Steyn. “It's about demonstrating real sector leadership so that the target companies they're engaging with say: ‘Yes, these are the partners we want for the future of the business.’”

The next generation city is here

Data centres defy easy classification: they’re part real estate, part infrastructure and part technology. Their future will therefore be shaped not by any single sector, but by the convergence of many, with power, capital, talent and trust at the heart of it all. The choices firms make today – where to invest, how to operate and whom to partner with – will decide who leads the AI economy and who falls behind.

Whether you are an investor, operator, technology provider or policymaker, our team is ready to help you navigate the evolving landscape of digital infrastructure and AI-driven growth.