The Factories of the AI Age: How Data Centres Are Reshaping Electricity, Risk, and Regulation

AI’s explosive demand for computing power is redrawing the global map of electricity, capital, and digital dependency. Data centres are becoming one of the fastest-growing infrastructure classes — but their returns increasingly depend on how countries manage electricity costs, grid stability, policy design, and whether demand actually materializes in the face of rapid efficiency gains.

The Return of Industrial Power

Artificial Intelligence (AI) is transforming key trends in physical infrastructure, and data centres have become the factories of the AI age. Their expansion now shapes demand for electricity systems, water demand, and telecommunications infrastructure in addition to investment flows, and regulatory choices as directly as traditional industrial sectors once did. Every major AI system relies on intensive computational power, and that compute relies on stable, secure electricity.

Global electricity demand from data centre infrastructure is projected to nearly double by 2030, reaching approximately 950 TWh per year—roughly equivalent to Japan’s annual electricity use (IEA, 2025 (opens in a new tab)). As computation scales, electricity is becoming the defining input for intelligence — and the availability of affordable, stable power is a strategic advantage for economies and for data center providers. Thus, AI has accelerated data centres’ status as strategic commodity — and electricity as a strategic bottleneck.^[1]

The Electricity Imperative of AI Infrastructure

Data centres have shifted to become essential industrial assets that anchor national economic growth and innovation. Their economics are increasingly determined not by their technological sophistication, but by their relationship to electricity systems. Access to low-cost, reliable power now determines the viability of new facilities. Grid-connection times, cooling capacity, and the ability to secure long-term power-purchase agreements also shape investment feasibility.

McKinsey estimates that roughly $6.7 trillion in global capital will be required for AI-driven data-centre buildouts by 2030 (McKinsey, 2024 (opens in a new tab)). This makes data centres one of the fastest-growing sources of electricity demand in the world.

Table 1 — Projected Electricity Use from Data Centres (from 2024 to 2023)

Different regions are approaching growth differently, with US having quicker ramp up than Europe. These regional divergences are already influencing where capital flows — and how it is priced. The value of a data centre increasingly depends on where it’s located — risks such as grid stability, electricity cost, and permitting speed now shape investment returns as much as technology or scale.

The Global Risk Landscape for Data Centres

Data-centre risk is now defined by a complex set of physical, regulatory, and geopolitical variables. As AI build-out accelerates, data centres function not only as engines of digital growth but also as exposure points to system-level infrastructure risks (grid strain, electricity affordability, digital dependency).

The risk profile of data-centre investment spans six core dimensions:

• Electricity Price Volatility – fluctuations in wholesale electricity markets, long-term PPA exposure, and sensitivity to regional fuel-mix volatility.
• Demand Risk - uncertainty about long-term compute demand due to efficiency gains in models, algorithms, and hardware.
• Grid Stability and Capacity – local congestion, connection delays, and grid resilience affecting uptime and expansion potential.
• Policy and Regulatory Environment – evolving incentives, permitting frameworks, environmental-performance standards, and potential requirements to shield residential consumers from industrial load costs (see Box 1).
• Renewable Access and Co-location Potential – the availability of affordable, low-carbon power sources and the ability to site near generation assets. This is required to meet the dual goals of a consistent base load to mitigate access risk and renewables to meet net zero goals.
• Geopolitical and Supply-Chain Exposure – vulnerability to trade restrictions, component shortages, or regional conflicts impacting construction and operations.

Understanding how these risks interact is essential. A market with low industrial electricity prices but high grid congestion may be less attractive than one with higher prices but predictable interconnection pathways. Similarly, strong renewable-access markets may reduce volatility, but policy selectivity can still limit the pace at which capital can be deployed.

Table 2 — Comparative Risk Landscape

Risk Implications for Investors and Policymakers

A bifurcation is emerging in the cost of electricity for data-centre operators. Facilities with secured grid access, long-term renewable PPAs, and predictable permitting benefit from structurally lower operating costs and lower cost of capital. In contrast, assets exposed to wholesale-market swings, congested networks, or uncertain planning processes face meaningfully higher true CoE.

For investors, this divergence directly influences valuations. Discount rates, leverage levels, and sensitivity to future electricity prices must be calibrated to local grid reliability and policy dynamics. For policymakers, balancing AI-driven load growth with consumer protection is challenging. Regions that reform interconnection processes and integrate AI-related demand into long-term grid planning will be more attractive destinations for capital. Conversely, regions that fail to adapt risk missing out on a critical period of digital infrastructure investment, especially if they do not provide alternative benefits such as tax benefits, free connection or otherwise.

The Repricing of Risk and Return

Traditional valuation frameworks based on fixed input assumptions no longer reflect the reality of data centre investment. Electricity-price volatility, policy uncertainty, and grid constraints introduce fundamentally non-linear risk. Probabilistic modelling, scenario analysis, and dynamic valuation approaches are required to capture the new distribution of risks and outcomes.

At Vallorii, we integrate forward-looking electricity trajectories, grid reliability distributions, policy-variability models, renewable-access pathways, and geopolitical scenarios into valuation systems. This allows both investors and policymakers to understand not only expected returns but also resiliency under stress. Such intelligence is no longer optional; it is central to underwriting the next wave of AI infrastructure.

Conclusion — The Intelligent Power Economy

AI is making electricity the currency of computation. Data centres now sit at the intersection of electricity policy, capital allocation, and industrial strategy. Their economics are determined as much by power availability and regulatory clarity as by the hardware they contain. The next decade will favour those who understand how physical and digital systems converge and create feedbacks—how electricity, grids, and policy shape the value and viability of AI.

Their economics are determined as much by power availability and regulatory clarity as by the hardware they contain.

Investors who integrate probabilistic modelling and real-time intelligence into their valuation processes will allocate capital more effectively and build more resilient portfolios. Intelligent capital will not only price risk more accurately — it will align the growth of AI with sustainable electricity and investment outcomes, but only if those risks are modelled and integrated into every valuation.