Governments around the world are racing to position themselves as leaders in AI. Policy announcements, funding packages and national strategies are coming thick and fast, all aimed at capturing a share of what is widely seen as the next industrial revolution.
In March the UK announced a national plan to build the infrastructure and capacity needed to power AI, innovation and economic growth across the country. The US has an action plan to achieve ‘global dominance.’
The Japanese AI Promotion Act, fully in force since September 2025, aims to position Japan as the ‘world’s most AI-friendly country.’
Article continues below
CEO & founder of CUDO Compute.
But there is a disconnect at the heart of many of these efforts. AI is still being treated primarily as a software or innovation challenge, when in reality it is increasingly constrained by physical infrastructure.
The ability to train, fine tune and run models at scale depends on access to compute, which in turn depends on land, power, and the speed at which infrastructure can be deployed.
That shift is already influencing where investment flows. Companies are making decisions based not on where AI policy sounds strongest, but on where infrastructure is available, affordable and operational.
If governments want to attract long-term AI investment, they need to focus less on abstract ambition and more on the conditions that allow deployment to happen in practice – land, power, and compute.
Treat energy as AI infrastructure
Energy is not a side conversation. It is the foundation of AI. Governments need to treat energy as core AI infrastructure, aligning power generation, grid access and data center development into a single, coordinated strategy that enables fast, reliable and cost-effective deployment.
AI workloads are energy intensive and highly sensitive to both cost and availability. By 2030, AI data centers could account for 8–12% of total US electricity demand, rising from around 3–4% today. When power strategy, grid access and data center development sit in different silos, projects stall before they begin.
This is already visible in the UK, where our research found that for 24.5% of tech organizations energy prices take over more than a third of AI infrastructure budgets, and 37% say that energy price volatility is increasing uncertainty when it comes to AI development.
If energy cannot be secured reliably and economically, investment will follow markets where it can.
Make speed a competitive advantage
In AI infrastructure, time directly affects viability.
Hardware cycles move quickly, and delays of 12 to 18 months can render initial investment assumptions obsolete. Many organizations can secure GPUs on paper, but struggle to bring them into live, production-ready environments due to bottlenecks in planning, permitting and grid access.
Take Ireland, where a surge in data center demand led to grid constraints and planning delays around Dublin, forcing the government and grid operator to effectively pause new connections in certain areas.
Projects that had secured land and investment were left waiting years for power access, showing how bottlenecks in permitting and grid capacity can stall deployment even when demand and capital are in place.
Governments that reduce friction in these processes and shorten the path from approval to activation will have a clear advantage. Access to infrastructure is no longer enough. What matters is how quickly it can be brought online.
Prioritize access to compute, not just innovation
There is still a strong focus on funding AI research, startups and applications. But without access to compute, those investments cannot scale. In Europe, initiatives like Horizon Europe have poured billions into AI research and innovation.
But access to high-performance compute remains fragmented across member states, with limited large-scale infrastructure compared to the US. As a result, many AI companies still rely on non-European cloud providers to scale.
This creates a growing tension between sovereignty and economics. In the UK, 45% of organizations say data sovereignty and regulation shape their deployment strategy, while 43% still prioritize cost and performance.
In practice, this means that if local infrastructure is too expensive or too slow to deploy, workloads move elsewhere. Attracting investment requires ensuring that capacity exists where organizations actually need it.
Enable regional, distributed infrastructure
AI infrastructure cannot be treated as something that can be concentrated into a handful of locations.
Deployment is governed by practical constraints such as land availability, existing power generation and grid connectivity. The fastest way to scale is to build in regions where these conditions already exist, rather than forcing development into zones that require significant new infrastructure before they can support meaningful capacity.
A more distributed approach not only accelerates deployment, it also reduces pressure on national grids and supports more resilient, scalable growth.
We have a data center in Bournemouth, UK; Eastern Europe is emerging as an attractive location – more than half of Romania’s energy mix comes from hydro and nuclear sources, power costs sit up to forty per cent lower than many Western European markets, and the country has a growing AI ecosystem with one hundred and forty six AI companies including $35 billion unicorn UiPath.
Align sovereignty, economics and skills with reality
Sovereignty goals, cost pressures and skills shortages are often treated as separate issues. In reality, they converge at the point of deployment.
For regulated industries, local compute is essential, but it must also be commercially viable and supported by the right operational expertise. The industry is facing a shortage not just of AI developers, but of engineers who can build and run large-scale infrastructure.
That means engineers who understand power, cooling, networking and hardware at scale, alongside platform and systems teams who can keep environments stable, efficient and secure. These are not skills you can retrofit once the data center is built, they need to be developed in parallel with infrastructure investment.
Without the skills to operate these environments, investment cannot be translated into reliable, long-term capability. Governments that align policy, economics and workforce development around deployment will be better positioned to support sustained AI growth.
Execute compute
The common thread across all five areas is execution.
AI investment is no longer driven by who has the most ambitious strategy or loftiest goals. It is driven by who can provide infrastructure that works, at scale, in the real world.
The countries that align energy, planning, compute access and skills around deployment will attract the next wave of AI investment. The rest risk becoming consumers of AI built elsewhere, rather than contributors to its development.
We’ve featured the best AI tools.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
https://cdn.mos.cms.futurecdn.net/U76sZeRd6fS2fKt5RqBYPL-2560-80.jpg
Source link
