December 2025 · Op-ed

The Next Nordic Export Industry

Compute, energy, and the next generation of industrial infrastructure. Written December 2025 — the public version of the thesis this site is built around, seven months before the site.

For much of its modern history, Nordic competitiveness has been built around abundant raw materials and in factories.

Not always factories in the narrow sense, but industrial systems that took abundant energy, raw materials, engineering capability, and long-term capital and converted them into products and infrastructure the world depended on. Hydropower, iron ore, timber, paper, electricity grid components, robotics, telecommunications, and now gaming were not developed primarily for domestic consumption. They achieved scale because they were designed for a global market, and because we learned repeatedly how to turn natural advantages into globally competitive industrial assets.

We find ourselves in a similar moment again.

Once more, competitiveness is increasingly shaped by electrons. Not only by how much electricity we generate, but by how effectively we convert it into economic work. In the 21st century, that work is increasingly computational, measured not in tonnes or units shipped, but in compute and tokens. What has changed is not the logic of value creation, but the form of the factory.

The next generation of factories is digital, capital-intensive, and tightly coupled to energy systems. They operate continuously, at scale, and over long asset lives. Like the industrial systems that came before them, they reward those who design for durability, efficiency, and system integration rather than short-term optimisation.

Data centres sit squarely in this lineage.

They are often discussed narrowly as land-use challenges or as large electricity consumers. Those concerns are real and should not be dismissed. But viewed only through that lens, data centres appear as problems to be managed rather than as industrial assets that shape long-term competitiveness.

This broader framing is increasingly reflected in European energy and policy analysis. Institutions such as the International Energy Agency have begun to treat data centres as structural components of future power systems and industrial development, highlighting both the opportunities they create and the risks associated with poor coordination, slow permitting, and grid constraints.¹ The diagnosis is sound. What now matters is how regions with strong industrial instincts respond.

For the Nordics, with abundant fossil-free power, capable utilities, strong institutions, and an engineering tradition built around exports, this should feel like familiar terrain rather than unexplored ground.

A familiar export logic, in a new form

Nordic prosperity has never been built on the assumption that domestic demand alone would carry growth. Export orientation has been a necessity, not a choice. What distinguished successful periods was not scale for its own sake, but the ability to combine natural resources, engineering, and capital into systems the rest of the world depended on.

Compute infrastructure increasingly fits this same pattern, and the emergence of token-based compute economics is an important step in making this industrial logic explicit.

Modern data centres, particularly those designed for advanced compute and AI workloads, convert electricity into digital output that can be sold globally. Across the technology stack, this output is now commonly measured in tokens, the unit of work produced by modern compute systems. In practical terms, data centres are no longer merely supporting infrastructure. They are production assets. Like any factory, they take a primary input, in this case electricity, combine it with capital, engineering, and operating discipline, and produce a standardised output that can be sold globally. The fact that the output is digital rather than physical does not change the industrial logic.

The parallel with earlier Nordic industries is instructive. Telecommunications infrastructure, for example, required sustained, capital-intensive research and investment over decades. Companies such as Ericsson and Nokia built systems well ahead of clear demand signals, and in doing so helped drive global standardisation that became the bedrock for much of the digital economy we see today, including the emergence of regional champions such as Spotify.

Compute infrastructure plays a similar enabling role today. It underpins advances in engineering, medicine, energy systems, logistics, defence, and industrial optimisation. Framing it primarily as a local power-pricing or land-use issue risks underestimating its strategic importance.

Scale matters, and domestic demand is not enough

One reality deserves to be acknowledged clearly. The Nordic region is too small to anchor globally competitive compute infrastructure on domestic demand alone.

The economics of next-generation compute increasingly favour very large installations, often hundreds of megawatts or even gigawatts, designed for high density, long asset lives, and continuous utilisation across multiple technology generations. These facilities can serve local workloads, but they are only competitive when designed for regional and global scale. Without them, many strategic workloads will be run elsewhere or at a higher cost, or worse, not at all.

This is not a departure from Nordic economic history. It is a continuation of it.

Latency constraints, once central to location decisions, matter less for many modern workloads, including AI training, inference, simulation, and industrial optimisation. This is partly due to their inherent characteristics and partly due to distributed edge infrastructure. What matters instead is access to reliable, affordable, low-carbon power, predictable timelines, and the ability to design assets that remain competitive over long horizons. On these dimensions, the Nordics remain unusually well positioned.

The implication is straightforward. If large-scale compute infrastructure is to be competitive, it must be conceived from the outset as export-oriented and built for global scale, much like the industries that built Nordic prosperity in earlier eras.

Competitiveness, sovereignty, and access to compute

European discussions of digital sovereignty often focus on regulation, ownership structures, and standards. These are important considerations. But sovereignty increasingly has a more practical dimension, namely access to compute capacity at scale and at competitive cost. These perspectives are increasingly visible in EU policy introduced in 2024 and 2025 and should be allowed to permeate national policy as well.

Scientific research, advanced manufacturing, pharmaceutical development, climate modelling, and defence applications are all becoming more compute-intensive. The next generation of digitally native companies, gaming studios, foundation model developers, and other emerging sectors cannot operate without it. Progress in these fields depends not only on talent and ideas, but on proximity to, or reliable access to, large-scale computing resources. If they cannot find it in Europe, they will go elsewhere or never emerge in the first place.

The Nordics have consistently punched above their weight at the application layer of software and AI. What is less certain is whether this advantage can be sustained without stronger anchoring in the infrastructure layer. Reliance on compute capacity located elsewhere introduces cost, capacity, and geopolitical dependencies that are often invisible until they become binding.

Hosting large-scale compute infrastructure domestically does not imply isolation or protectionism. On the contrary, it is an opportunity to build the next generation of export economy. It is about ensuring that Nordic companies, researchers, and public institutions are not structurally disadvantaged as compute becomes a core input across the economy.

Why waiting carries risk

Given the scale of investment and the pace of technological change, a cautious posture is understandable. Infrastructure decisions of this magnitude are inherently uncertain.

Yet recent history suggests an important asymmetry. The cost of waiting can be as real as the cost of acting too early.

Over the past decade, the build-out of compute and data centre infrastructure has repeatedly exceeded earlier expectations. As new workloads, from cloud services to AI, have emerged faster than anticipated, projections have been revised upward rather than downward. The International Energy Agency now estimates that electricity demand from data centres worldwide will more than double by 2030, driven in large part by high-density compute and AI workloads.² Similarly, McKinsey analysis points to sustained double-digit annual growth in demand for AI-ready data centre capacity through the end of the decade.³

These trends do not eliminate risk. But they underline a structural challenge. Grid upgrades, permitting, and site development require long lead times. Under-building, short-term design choices, or delayed coordination can lock regions into sub-scale or poorly integrated assets that are difficult to correct later, or cause assets to become obsolete sooner than necessary.

Design and coordination matter more than scale alone

The IEA is right to emphasise that energy systems must keep pace with data centre expansion. But scale itself is not the core problem. Design and coordination are.

Poorly integrated projects, designed around near-term assumptions and connected late to grid planning, can strain local systems and provoke public opposition. By contrast, infrastructure designed from the outset for long lifetimes, high flexibility, and system integration can support grid stability, improve utilisation, and in some cases reduce overall system costs.

This is where Nordic institutions have a demonstrated advantage. Strong utilities, capable system operators, trusted regulators, and a tradition of long-term planning make it possible to approach data centre development as a systems problem rather than as a sequence of isolated connection requests.

Moving in this direction does not require centralised planning. It does require a shift from reactive responses to intentional coordination across power generation, grid investment, permitting, site design, construction, and operations.

From diagnosis to action

Europe's policy debate has largely converged on the diagnosis. Energy constraints risk slowing the expansion of data centre capacity needed to support digital and AI ambitions. The next step is to move from diagnosis to execution.

For the Nordics, this means recognising compute infrastructure as an extension of a long-standing economic model. It means converting energy and engineering into exportable value while safeguarding system stability and consumer interests. The starting position is strong. What matters now is whether this advantage is used deliberately.

Electrons powered the industries that built Nordic prosperity. Increasingly, they will also power the infrastructure that converts ideas, engineering, and science into global economic value. The transition is already underway. The remaining question is whether it will be shaped proactively or whether its contours will be defined elsewhere and accommodated only after the fact.

Sources

— Oscar · December 2025