From Code to Concrete: How AI and Geopolitics are Powering the Infrastructure Cycle

Have you noticed the shift? The AI story used to be about software. Now it has become industrial.  What began as a conversation about software valuations has moved to power plants, data centers, and physical real estate. For long-term investors, this distinction matters.

AI requires enormous amounts of electricity, physical space, and high-speed connectivity. These inputs must be built, maintained, and expanded. Compute may feel virtual, but the underlying infrastructure is stubbornly real.

Recent geopolitical tensions have only sharpened that focus. When a narrow stretch of water becomes the central headline across the globe, every country is quickly reminded of the importance of economic sovereignty. And when systems appear fragile, governments do not respond with optimism. They often respond with spending, coordination, and a renewed appetite for controlled redundancy.

This is a shift worth monitoring. AI is accelerating demand for electricity and connectivity, while geopolitics is simultaneously raising the value of reliability. Together, these forces are bringing infrastructure to the forefront as a potentially defining investment theme of this decade.

The AI trade is maturing

AI’s next chapter is becoming less constrained by imagination and more by delivery. In practical terms, that means waiting years for electrical grid connections, sourcing specialized equipment that can take years to manufacture, and navigating complex permitting processes. While market debate continues around potential winners, there is little debate regarding the need for power, space, and bandwidth to operate AI models.

The AI narrative has started to split into two different debates. One debate centres around economics: user base, pricing, and durable competitive advantages. The other debate is about physical constraints: how quickly new data center capacity can be deployed to support the training and delivery of new models, factoring in power, land, and connectivity constraints.

For investors, the key insight is: it may not matter which AI company ultimately wins the software race. More predictable is that any successful player will require substantial, long-lived physical infrastructure to operate. The companies supplying power, land, and connectivity may benefit from durable demand for years to come, regardless of how competitive landscapes evolve.

Electricity demand scales with its availability

The scale of data center electricity demand is now large enough to shape national planning. Concerns around energy use, water demand, and local planning have already driven 11 U.S. states to propose temporary bans on new data center construction.

A typical AI-focused data center can consume as much electricity as 100,000 households, with some of the largest AI data centers currently under construction expected to consume roughly 20 times that. The International Energy Agency (IEA) projects that global data center power consumption could more than double by 2030.  

In addition to securing reliable power, data center campuses require new real estate development, network interconnection, and investment into the GPUs themselves. The global investment services firm JLL estimates that meeting expected global demand could require up to USD$3 trillion of investment by 2030, including roughly USD$1.2 trillion of real estate assets. 

At Nicola Wealth, we have invested across the underlying physical infrastructure that supports existing data demand, with potential upside benefit from AI demand growth. For example, Nicola Global Infrastructure Limited Partnership (NGI) is invested in a portfolio of hyperscale data centers in Paris, Madrid, and Milan, where power, land, and permitting constraints create high barriers to entry. These data centers have long-term contracts with hyper-scalers, supporting stable predictable cashflows. NGI has a 26% allocation to digital assets as of March 31, 2026, with global investments across data centers, fibre networks, and mobile network towers. NGI is also invested in the supporting power infrastructure with a 17% allocation to power and renewables as of March 31, 2026.

Geopolitics is reshaping how efficiency is valued

Modern supply chains were designed for a relatively predictable world. If a component could be made cheaper halfway across the planet, it was. If inventory could be trimmed, it was. Efficiency became the dominant operating principle, and it worked well until the recent Iran war reminded everyone that global trade still moves through narrow places.

The Strait of Hormuz provides a clear case study on the fragility of modern supply chains. It is one of the world’s most critical oil transit chokepoints, with roughly 20 million barrels per day of crude and oil products transiting in 2025. This represents approximately one quarter of the entire world’s seaborne oil trade.

There are limited alternatives; the IEA estimates that only 3.5 to 5.5 million barrels per day of spare pipeline capacity can bypass Hormuz through existing routes in Saudi Arabia and the United Arab Emirates (UAE). This leaves an implied shortfall that cannot be easily addressed with logistics alone. For Canadian investors and consumers, this is not an abstract concern. Energy price shocks can transmit quickly through the broader economy.

Consumers are already experiencing rising gas prices at the pump with effects cascading throughout the wider economy. Shipping, fertilizer, plastics, and anything delivered “just in time” are also seeing cost increases.

When markets seize, governments tend to spend

Governments have taken notice and are doubling down on sovereign independence.   There are numerous examples of governments underwriting infrastructure because they view it as a strategic necessity.

In Europe, the rationale has become explicit: submarine communication cables carry 99% of intercontinental internet traffic, making economic infrastructure extremely vulnerable to foreign sabotage. The European Commission has earmarked €347 million to strengthen cable security and repair capacity before the next incident forces the issue.

The U.S. reflects a similar dynamic through different channels. The Department of Energy’s Winter 2025 infrastructure progress report indicates that DOE has committed over USD$170 billion for grants and loans through major federal packages. In grid-specific terms, these investments are expected to enable 67 GW of grid capacity and new resources, and support the build-out and upgrade of more than 4,375 miles of new transmission lines by 2031. For context: 4,375 miles is approximately 18% the length of the Earth’s circumference.

Here in Canada, the 2025 federal budget committed CAD$115 billion in new infrastructure investment over the next five years. Critical minerals have also become a sovereign concern. In 2024, Canada reported 171 advanced critical mineral projects, in addition to CAD$2.1 billion of exploration.

Given current geopolitical instability, it seems only logical that the trend of critical infrastructure expansion will continue. At Nicola, we see compelling opportunities in the energy transition sector, as energy transition has become synonymous with national energy security. NGI has a 14% allocation to energy transition as of March 31, 2026.

The scale extends beyond any single narrative

AI is one accelerant. Geopolitics is another. However, the underlying infrastructure gap predates both trends.

Oxford Economics, in collaboration with the Global Infrastructure Hub, estimated that global infrastructure needs $94 trillion invested between 2016 and 2040; about $3.7 trillion per year. Meeting this need would require global infrastructure investment to increase to about 3.5% of annual global GDP; $618 billion more per year than current spending patterns. In practical terms, this implies the world needs to spend more annually on infrastructure than Apple’s annual gross revenue, and to sustain that rate for a generation.

We are not watching a theme. We are watching a system catching up with its own maintenance and expansion requirements.

Investing in bottlenecks

As AI becomes more deeply embedded in the economy, bottlenecks shift from code to concrete. While software features can be deployed quickly, physical assets such as a new substation, transmission line, or fiber route require longer development timelines.

At the same time, as geopolitical pressures intensify, governments are increasingly prioritizing economic sovereignty and capacity. That investment tends to concentrate capital in assets that are slow to replicate and hard to substitute.

In this context, infrastructure becomes more than asset allocation. It represents a way to express a view on persistent demand for resources, connectivity, and national resilience.

Within a diversified portfolio, infrastructure assets have historically played a distinct role from traditional assets like equities or fixed income. Infrastructure assets tend to generate relatively stable, often inflation-insensitive cash flows, frequently supported by long term contracts of regulated return frameworks. The portfolio case is less about massive growth and more about durability. It's about owning assets aligned with demand that is unlikely to disappear, regardless of which specific technologies or companies ultimately prevail.

For investors, the key point is that infrastructure does not need to guess which AI model will win. The more useful question is whether the world will keep needing resources and connectivity. On that, the answer looks unusually clear.