The world is in the middle of a defining shift in how it produces, moves and consumes energy. Fossil fuel dependence has re-emerged as a source of geopolitical fragility, electricity demand is outpacing supply and the cost of energy has become a central concern for consumers, companies, and governments alike.
The lesson from the current Middle East crisis is the same one history keeps repeating: dependence on fossil fuels concentrated in politically volatile regions exposes us to risks that no strategic reserves can fully insulate against. Diversifying energy sources is an essential part of the answer, but building that new capacity takes time. In the meantime, one truth remains as relevant as ever – the cheapest energy is the one we do not use.
The investment case for efficiency is stronger than it has been in decades, which is why resource efficiency represents our largest portfolio allocation at about 65%. The products and technologies behind this allocation are not abstract bets on a distant future: they are already deployed, saving energy and delivering measurable results.
While our portfolio remains deliberately diversified – across themes, return drivers and sources of alpha – two themes are currently standing out with particular strength. Each is underpinned by structural forces we believe will drive sustained opportunity over the long term.
Efficiency takes centre stage in AI
The AI race began as a race for raw computing power but has since quietly transformed into something more fundamental – an energy race. The grid simply cannot be built out fast enough or cheaply enough to keep pace with demand. Data centres already consume roughly as much electricity as Germany today – and will consume twice as much by 2030.
For the biggest players in AI, energy efficiency has become a business-critical priority, and the companies that can help them use less energy to do more are increasingly sitting at the heart of the AI opportunity.
For the past few years, the spotlight has been on NVIDIA, the dominant player supplying the powerful but energy-hungry GPUs (Graphics Processing Units) that train AI models. But GPUs are built for flexibility and raw power, not efficiency. As AI becomes used at scale – with billions of queries processed every day – a different kind of chip is emerging as critical: the ASIC (Application-Specific Integrated Circuit).
ASICs can deliver significantly more computing power per unit of energy, directly addressing the efficiency challenge that is now at the top of every hyperscaler’s agenda. We hold Broadcom, Marvell Technology and Alchip, which are among the biggest players in this space.
The search for efficiency is not limited to chip design. It is broadening to the overall hardware and software ecosystem that can deliver efficiency gains. Within the value chain, we also see compelling opportunities in networking, and the high-speed connections essential to keeping data moving quickly and efficiently – such as Arista, Accton and Fastly.
In addition, building more efficient chips requires highly advanced machines that only a handful of companies worldwide can produce – like ASML, ASM International and Lam Research.
The grid struggles to feed demand
For the past two decades, grid investment has been treated as routine maintenance rather than as strategic infrastructure. That era is over. A combination of AI-driven data centre expansion, the electrification of transport and heating, and the reshoring of manufacturing is reshaping the load curve in a way the system was never designed to handle.
Globally, power demand is projected to grow at a 3.6% cyclically-adjusted growth rate (CAGR) between 2026 and 2030 – a pace 50% faster than the previous decade. The symptoms are clearly visible in prices. The key question is not whether demand for compute power slows – it is how quickly infrastructure modernisation and build-out can catch up.
This bottleneck translates into one of the most durable capex cycles of the coming decade. Global grid spending has risen from $300bn in 2020 to $480bn in 2025, with a further $5.8trn forecast for 2026-2035. The US alone is absorbing roughly $1trn.
Our aim is to capture this spend along the full value chain of grid modernisation and build-out. These are the specialty contractors physically upgrading transmission networks, such as MasTec and Quanta Services, and the cable and equipment manufacturers supplying high-voltage cables, like Prysmian.
We also have exposure to transformers and electrical components that are in acute global short supply through WEG, as well as to regulated utilities deploying capital into transmission and generation on both sides of the Atlantic, including names such as National Grid and NextEra.
Ultimately, the grid is where the AI boom, the energy transition and the reshoring of industry all physically collide. Resolving this bottleneck is not optional – it is a prerequisite for almost every major growth theme of the coming decade. That is what makes grid modernisation such a powerful and durable investment opportunity.
Thomas Sørensen is portfolio manager of Nordea’s Global Climate and Environment strategy. The views expressed above should not be taken as investment advice.