Every AI chip eventually needs somewhere to plug in. That connection point has become one of the tightest constraints in the entire AI build-out. More from EasyAssetManagement.
Part of our thematic investing series, following on from our webinar and our recent blogs on commodities and semiconductors.
At EasyAssetManagement, thematic investing starts with understanding the entire ecosystem behind a structural trend, not just the companies making the headlines.
Artificial intelligence is a good example. Many investors think of AI as a handful of software companies or chipmakers. We see it as a much broader ecosystem. It begins with the raw materials used to manufacture semiconductors, extends through the equipment used to produce them, the companies that design and package the chips, and the industrial businesses building, connecting and cooling the data centres where those chips ultimately operate.
Looking at AI through this lens reveals that every part of the value chain matters. It also highlights where the real constraints lie. One of the most important sub-themes within AI is energy. Every AI data centre requires a reliable supply of power, making electricity not just another input, but one of the fundamental enablers of the entire AI ecosystem.
That naturally led us beyond semiconductors and into power infrastructure. What first appeared to be a supporting industry has become a compelling AI investment theme in its own right, benefiting from the same spending boom and build-out driving the cutting edge of AI.
The International Energy Agency (IEA) projects that electricity used by data centres worldwide will roughly double between 2025 and 2030, to around 950 terawatt hours, slightly more than Japan's entire annual consumption today. A single advanced AI server rack, roughly the size of a large fridge, is on track to draw as much peak power as 65 households by 2027.
Source: IEA (2026), Electricity consumption by data centres, 2020-2035, IEA, Paris
Everyone talks about AI in the abstract. The electricity bill is where it gets concrete, and the United States is at the centre of it.
Data centres have become one of the biggest new sources of electricity demand in the country. After years in which power consumption remained stable, demand is rising again, and the computing boom is a large part of why.
The catch is that electricity supply cannot keep pace. A data centre can often be built far more quickly than the power infrastructure and grid connections needed to run it. In many of the busiest markets, grid connection queues already stretch for years. Building new generation and connecting it to the grid is a slow process, constrained by permitting, planning and lengthy interconnection queues. As a result, the bottleneck is increasingly not the construction or fitting out of the data centre itself, but securing the electricity needed to power it. That is becoming a meaningful constraint on how quickly new AI capacity can come online.
When demand races ahead of supply, the important question for investors is not whether a shortage exists. It is who owns the scarce capacity that everyone suddenly needs.
One of the quickest ways to add large-scale, always-on power in the US is through a natural gas plant. At the centre of every gas plant is a heavy-duty gas turbine.
Three manufacturers dominate this market globally: GE Vernova, Siemens Energy and Mitsubishi Power.
All three are effectively sold out.
GE Vernova, a current holding in both our funds, has a record gas turbine backlog, and management has indicated that production slots are reserved years into the future.
Siemens Energy and Mitsubishi Power are reporting a similar picture. Their order books are at or around record levels, while new delivery slots extend well into the second half of the decade.
Pricing on new reservations has risen accordingly.
This is what a supply constraint looks like on a company's income statement. Manufacturers operating in a mature and traditionally unglamorous industry suddenly have multi-year demand visibility and greater pricing power.
This is where the thematic process becomes particularly valuable.
When an important machine is sold out, we ask the same question we explored in our semiconductor blog, what can that machine not be built without?
For gas turbines, part of the answer lies in the specialised components and materials.
Turbines are built using specialised components, metals and alloys that cannot easily be substituted. They are chosen for their ability to withstand extreme heat, pressure and mechanical stress, while producing them to the required specifications is a complex and highly specialised process.
The turbine manufacturers attract most of the attention. The material and alloy suppliers sit one layer beneath them, exposed to the same scarcity but often receive far less attention from investors.
Source: Constellation
Natural gas is one of the quickest answers to the immediate power shortage. But the map must also consider where reliable, carbon-free electricity will come from over several decades.
That points towards nuclear power.
The shift over the past two years has been remarkable, and it is increasingly being driven by the technology companies themselves.
The hyperscalers, the small group of companies building the world's largest AI data centres, are now pursuing nuclear power at scale. They are signing long-term electricity supply agreements with existing plants, supporting the restart of retired reactors and investing in new nuclear projects. This includes small modular reactors, which are designed to be manufactured in factories rather than constructed entirely on site.
Governments are moving in the same direction, providing policy and funding support for new reactor programmes in the US and elsewhere.
Every nuclear reactor requires fuel, equipment and specialised components. This led us towards the wider nuclear supply chain, rather than relying on the success of any single project.
Our Global Equity fund holds Cameco, one of the world's largest uranium producers and the owner of 49 per cent of reactor builder Westinghouse. It also holds BWX Technologies, which manufactures nuclear reactor components and nuclear fuel.
Some AI projects cannot afford to wait for a gas turbine production slot or a grid connection. This has pushed many data centres towards behind-the-meter solutions that generate power directly on site.
That creates an opportunity for fast, on-site power solutions. One of the most visible beneficiaries has been Bloom Energy, a holding in both our funds.
Bloom's fuel cells convert natural gas into electricity without combustion. Its systems are factory-built and can be deployed far more quickly than a traditional grid connection can be secured.
Oracle has contracted gigawatt-scale capacity from Bloom, including for a New Mexico AI campus which is designed to operate essentially off-grid.
The market has taken notice, and a significant amount of future growth is now reflected in Bloom's share price.
Securing enough electricity generation capacity is only the first step. That power still needs to be connected to the data centre, moved across the grid, transformed to the correct voltage, distributed through the facility and converted into the precise form required. At the same time, the enormous amount of heat produced by those chips must be removed to keep the system running safely and efficiently.
Following the power therefore leads naturally from generation into transmission, electrical equipment, construction, power management and cooling. Each is part of the same physical system, and each must expand if AI infrastructure is to grow at the pace currently expected.
Consider the journey.
We began with one long-term structural trend: artificial intelligence. By mapping the theme all the way through to its physical limits, we uncovered an entire investment opportunity set.
This includes turbine manufacturers with record backlogs, specialist material and alloy suppliers, the nuclear fuel chain, on-site power providers, and the builders and equipment companies supporting the entire system. We invest across many of these areas through our thematic strategies, giving us exposure to different parts of the same underlying structural trend.
None of these investments required us to predict which chatbot would eventually win.
They required us to follow the theme to the points where growing demand meets limited supply, and then carry out detailed, bottom-up research on each business. That means assessing the durability of demand, profit margins, balance-sheet strength and valuation.
Understanding the entire AI value chain and the AI energy sub-theme is central to our investment process. It helps us identify the businesses we believe are best positioned to benefit from the theme and informs our AI exposure across our global equity strategies, including the EasyETFs AI World Actively Managed ETF, EasyETFs Global Equity Actively Managed ETF, and the offshore equity allocation within the Regulation 28-compliant EasyETFs Balanced Actively Managed ETF.
If you are looking for exposure to global equities, AI-themed opportunities, or a balanced investment strategy, check out our EasyETFs Global Equity Actively Managed ETF, EasyETFs AI World Actively Managed ETF and EasyETFs Balanced Actively Managed ETF.
Any opinions, news, research, reports, analyses, prices, or other information contained within this research is provided by an external contributor as general market commentary and does not constitute investment advice for the purposes of the Financial Advisory and Intermediary Services Act, 2002. First World Trader (Pty) Ltd t/a EasyEquities (“EasyEquities”) does not warrant the correctness, accuracy, timeliness, reliability or completeness of any information (i) contained within this research and (ii) received from third party data providers. You must rely solely upon your own judgment in all aspects of your investment and/or trading decisions and all investments and/or trades are made at your own risk. EasyEquities (including any of their employees) will not accept any liability for any direct or indirect loss or damage, including without limitation, any loss of profit, which may arise directly or indirectly from use of or reliance on the market commentary. The content contained within is subject to change at any time without notice