By Robert Swift

There is a familiar rhythm to modern markets. A new technology arrives, investors pile into the glamorous bit, and then, a little later, reality wanders in wearing steel caps and carrying an invoice.

AI has been no different. The first phase was about chips, models, and the usual Silicon Valley theatre, equal parts genius and PowerPoint. The second phase is proving rather less ethereal. Data centres need power. Vast amounts of it. Not just on paper, but physically delivered, stepped down, switched, protected, transmitted, cooled, backed up, and made reliable enough that a trillion-dollar software stack does not fall over because a transformer is late. The International Energy Agency now expects data centre electricity consumption to more than double to roughly 945 TWh by 2030, with AI the biggest driver. In the United States alone, data centres are expected to account for nearly half of electricity demand growth to 2030. At the same time, the IEA says grid capacity is already a critical bottleneck, with more than 2,500 GW of renewable, large-load, and storage projects stalled in grid queues worldwide. 

That is the real story. The AI boom is not just a computing story. It is an electricity story. More specifically, it is a story about the grubby, underloved, marvellously unglamorous assets that allow electrons to behave themselves. Transmission lines. Substations. Switchgear. Transformers. Cable systems. Grid automation. Power distribution architecture. Contractors who can actually build the thing. Every digital revolution, after enough champagne and TED Talk energy, discovers concrete, copper, steel, and lead times.

Markets are beginning to grasp this, but still not fully. The popular framing remains that AI is a race for compute. True enough, as far as it goes. But compute without power is just very expensive furniture. AI data centres are not merely large office buildings with better branding. The IEA notes that AI-focused data centres can draw as much electricity as power-intensive factories, but are far more geographically concentrated. That concentration matters, because it is precisely what turns a broad energy demand story into a local infrastructure problem. The bottleneck is not simply generating more power, it is getting enough high-quality power to the right place, at the right voltage, with acceptable reliability, and without waiting the better part of a political cycle.

That brings us to the picks and shovels. Not the AI darlings that dominate cocktail chatter, but the companies building the electrical spine of the whole affair. Here are five listed global infrastructure names that look well placed.

1. Eaton, when the power room matters more than the server room

If AI is the glamorous tenant, Eaton is the landlord’s electrical contractor, and that may turn out to be the better business. Eaton sits squarely in power management, and increasingly in the parts of the stack where data centre growth meets physical constraint. In its 2025 annual report, Eaton said it had announced $13 billion of acquisitions to expand its leadership in electrification and digital infrastructure, including Resilient Power’s medium-voltage solid-state transformer technology for high-density data centres and Fibrebond’s modular power enclosures for data centre and utility applications. It also described data centres and utilities as two of its most exciting segments, while its third quarter 2025 analyst presentation said electrical sector and Electrical Americas data centre orders were each up about 70% year on year, with revenue up about 40%.

That is not a fashion statement. It is operating evidence that the constraint is moving deeper into the electrical architecture. Eaton benefits because a data centre is not one thing. It is a stack of power quality, protection, distribution, backup, enclosure, and increasingly higher-density infrastructure challenges. As workloads intensify, the value shifts toward companies that can help manage power through the facility, not merely get it to the boundary fence. Eaton’s language about serving customers “from the grid to the data center” is telling. That breadth matters in a market where delays are costly and integration risk is unpopular.

The investment case, then, is not that Eaton is some secret AI stock. Quite the opposite. It is better than that. It is a quality industrial sitting inside a structural capex cycle, with real exposure to one of the most urgent pain points in the buildout. That tends to be a healthier place to be than the centre of the hype carnival.

2. Schneider Electric, the adult in the room

Schneider Electric has the sort of positioning that investors often ignore until they can no longer afford to. It lives in energy management, distributed power, building systems, software, and automation, which is a polite way of saying it sits where electricity, efficiency, and uptime have to make peace with each other. In Schneider’s half-year 2025 results, North American Energy Management grew 14.6% organically, with strong double-digit growth in systems led by the data centre segment. Services also posted strong momentum in data centres. In its third quarter 2025 release, Schneider said group revenue rose 9% organically, with Energy Management up 10%, led by continued strong growth in data centres. CEO Olivier Blum explicitly said sustained data centre demand drove performance, supported by investments in grid infrastructure.

This is the important distinction. Schneider is not just selling boxes. It is selling a system. AI infrastructure has a nasty habit of exposing every weakness in the chain, from switchboards to cooling to monitoring to service response. Hyperscalers want speed, but they also want standardisation, modularity, and energy visibility. Utilities want loads they can manage. Enterprise customers want reliability without needing a priest, an engineer, and a rescue helicopter every time demand spikes. Schneider plays across those layers.

There is also a subtler point. As the market finally internalises that electricity is not free, efficiency becomes part of capacity creation. An extra megawatt saved can be almost as useful as an extra megawatt supplied. Companies that improve distribution, controls, monitoring, and utilisation are not peripheral to the capex cycle, they are central to it. Schneider has that advantage of being both physically embedded and operationally relevant, which is another way of saying it gets paid while everyone else argues about the future.

3. Siemens Energy, because grids do not upgrade themselves

If one wanted a pure expression of the thesis that AI eventually runs into electrical reality, Siemens Energy would be hard to ignore. Its annual report says the strong growth of data centres is expected to lead to further increased investment in grid infrastructure, generation capacity, and energy-efficient technologies, with all of Siemens Energy’s business areas expected to benefit. More concretely, the company’s data centre solutions page lays out exactly what that means: HV substations, GIS and AIS switchgear, high-voltage transformers, grid stability tools, distribution transformers, and digital assets for grid connection and power quality. In fiscal 2025, Siemens Energy’s order backlog rose to a record €138 billion, Grid Technologies delivered its highest quarterly revenue to date, and Grid Technologies posted a 16% profit margin before special items for the year.

That is a formidable combination of relevance and momentum. It is also pleasingly literal. Investors love to say a company has “exposure” to a theme. Siemens Energy does not merely have exposure. It makes the things that stand between a giant data centre and a polite note from the utility saying, regrettably, not this decade. HV substations and transformers are not conceptual beneficiaries. They are the actual choke points.

There is a further attraction here. Grid spending is not solely an AI story. It is also about electrification, reliability, renewable integration, industrial load growth, and energy security. That means Siemens Energy is not hostage to a single narrative. AI helps accelerate the urgency, but the broader capex underpinning is diversified. In a world where many thematic investments are one-trick ponies dressed up as revolutions, that is rather refreshing.

4. Quanta Services, the people with the hard hats and the schedule

A great many investors prefer companies with lots of intellectual property and as few muddy boots as possible. That is aesthetically understandable and financially not always wise. Quanta Services matters because none of this electrical buildout happens just because an investor deck says it should. Someone has to engineer, install, connect, upgrade, and maintain the physical network. Quanta is one of the big contractors and service providers to electric utilities and infrastructure customers. Its investor materials say 70% of 2025 revenue came from Utility and Power, while its fourth quarter and full-year 2025 release reported record total backlog of $44.0 billion, reflecting accelerating demand in its Electric segment. A related company summary for year-end 2025 cited Electric segment backlog of $36.2 billion, also a record.

That is what demand looks like when it leaves the spreadsheet and enters the field. Quanta is a beneficiary of grid hardening, transmission expansion, substation work, utility modernisation, and increasingly the broader technology and communications ecosystem that sits alongside large-scale power infrastructure. Importantly, contractors with scale, customer relationships, labour availability, and project execution experience often become more valuable precisely when capacity is tight. In a rush, everyone wants the same people. There are not that many of them.

This is one of the more underappreciated truths in infrastructure investing. Bottlenecks do not just reward component manufacturers. They also reward the firms that can turn components into functioning assets. When lead times stretch and projects multiply, execution becomes scarce too. Scarcity, as markets occasionally rediscover, is usually good for returns.

5. Prysmian, because electricity and data both need a path

Prysmian is best known for cable, which sounds dull until one reflects that both power and data require transmission media, and civilisation has become unreasonably dependent on both. Prysmian’s 2025 reporting captures this neatly. The company said accelerating digitalisation, the rollout of data centres, and AI were supportive trends, and in 2025 it announced a partnership with Relativity Networks to co-manufacture hollow-core fibre and cable demanded by data centre operators in an AI-powered economy. Prysmian explicitly noted that the enormous demand for electricity to power AI-related data processing had created a potential bottleneck in new data centre construction, and argued this technology could help hyperscalers locate data centres closer to power sources. Separately, Prysmian also won a long-term agreement with Norway’s Statnett for extra-high-voltage underground cable systems, a reminder that the company is central to power-grid buildout as well as digital connectivity.

That dual exposure is interesting. The naïve AI trade says buy compute. The smarter infrastructure trade says buy the routes through which power and information must travel. Prysmian participates in both. It is not the hero of anyone’s science-fiction narrative, but it is highly relevant to the practical one.

There is also a geopolitical angle. Energy security is no longer an academic discussion in Europe or much of the developed world. Network resilience, domestic and regional grid investment, interconnection, and industrial sovereignty have all moved up the agenda. Cable may be mundane, but it is also mission critical. Markets usually pay up for mission critical things eventually. They just prefer to be dramatic about it first.

Why this theme still matters

The common thread across all five names is not simply “AI exposure”. That phrase is already becoming a little too available, like luxury in apartment brochures. The common thread is that each company sits at a point of genuine scarcity in the electricity and grid value chain. Eaton in power distribution and management, Schneider in energy management systems and services, Siemens Energy in substations and transformers, Quanta in actual construction and utility execution, and Prysmian in the cable backbone for both power and digital traffic.

The larger point is that AI capex is likely to prove far more infrastructure-heavy than many investors first assumed. That does not mean every infrastructure stock is a winner, nor that valuations cannot overshoot. They can and they will. But the direction of travel is increasingly hard to miss. If data centre power demand more than doubles by 2030, and if grid queues are already clogged, then the world will need far more spending on transmission, distribution, power quality, and connection infrastructure. This is not speculative futurism. It is industrial arithmetic.

In other words, the next leg of the AI story may be less about silicon magic and more about electrical plumbing. That may sound less exciting, which is precisely why it may still be investable.

TAMIM Takeaway

The market’s first instinct was to buy the brains of AI. The more durable opportunity may lie in funding its nervous system and power supply. Semiconductors matter, of course, but they are not much use without substations, transformers, switchgear, cable, and a grid that can cope. Eaton, Schneider Electric, Siemens Energy, Quanta Services, and Prysmian all sit in different parts of that bottleneck. None are especially romantic. That is part of the appeal. Real fortunes are often made not from owning the shiny object, but from owning the businesses the shiny object cannot function without.

________________________________________________________________________________________________

Disclaimer: Quanta Services (NYSE: PWR) is held in TAMIM Portfolios as at date of article publication. Holdings can change substantially at any given time.