{"id":51305,"date":"2025-11-03T09:41:15","date_gmt":"2025-11-03T14:41:15","guid":{"rendered":"https:\/\/www.cambridgeassociates.com\/?p=51305"},"modified":"2025-11-26T12:00:10","modified_gmt":"2025-11-26T17:00:10","slug":"vantagepoint-electrifying-returns-in-the-ai-era","status":"publish","type":"post","link":"https:\/\/www.cambridgeassociates.com\/en-as\/insight\/vantagepoint-electrifying-returns-in-the-ai-era\/","title":{"rendered":"VantagePoint: Electrifying Returns in the AI Era"},"content":{"rendered":"

Artificial intelligence (AI) is not just transforming business models and economic growth<\/a><\/strong>\u2014it is contributing to a step change in global electricity demand, with far-reaching effects on energy markets, utility infrastructure, and investment portfolios. This shift is creating both acute challenges and compelling opportunities. Grid reliability, infrastructure bottlenecks, and decarbonization pressures are intensifying, while the scale and urgency of required capital investment are unprecedented. Solutions that address these challenges\u2014modernizing the grid, expanding clean and reliable power, and improving efficiency\u2014offer the most appealing prospective returns.<\/p>\n

Yet, as with any technological revolution, uncertainty remains. The pace of AI adoption, the trajectory of energy efficiency, the availability of labor and materials along the electricity supply chain, and the ability of policy and capital to keep up with demand will shape the investment landscape. Investors must allocate capital thoughtfully, balancing optimism about the scale of the opportunity with discipline and selectivity.<\/p>\n

In this edition of VantagePoint, we examine how the rise of AI is reshaping the global energy landscape and highlight the most compelling opportunities and risks for investors. We analyze the drivers of surging electricity demand, the infrastructure and supply chain constraints that threaten to limit growth, and the comparative merits of emerging power solutions. We conclude with actionable investment implications and recommendations for navigating this rapidly evolving landscape. The winners will be those who can identify and invest in the infrastructure and technologies that enable the next era of digital and energy growth.<\/p>\n

The global demand surge<\/h2>\n

Hyperscale data centers powering advanced AI workloads are proliferating rapidly, as technology firms race to secure power contracts and build new capacity. This surge in AI-driven demand is layered atop powerful secular trends: digitalization, electrification of transport and industry, and manufacturing reshoring.<\/p>\n

Projections from the International Energy Agency (IEA) suggest global electricity demand could grow 1.5x to 3x by 2050 (2%\u20134% annually), with data centers alone expected to double their share of global consumption by 2030 to nearly 3% (Figure 1). In the United States, data centers may account for as much as 12% of total power demand by 2028, up from 4.4% in 2023.<\/p>\n

\"Stacked<\/p>\n

Historically, technology platform shifts have made demand forecasting difficult. For example, the move to cloud computing delivered dramatic efficiency gains that tempered earlier projections of data center power use. Today, advances in AI hardware, software, and cooling are again driving efficiency improvements, but the sheer scale and intensity of AI workloads mean that aggregate demand is likely to rise sharply. Further, should AI electricity demand fall short, other demand drivers should support investments in expanding electricity capacity.<\/p>\n

These forces are converging at a time when many power grids are already strained by aging infrastructure, intermittent renewable generation, and extreme weather events. The result is a widening gap between projected demand and available, reliable supply\u2014creating acute challenges for grid reliability and decarbonization.<\/p>\n

For investors, this environment presents a generational opportunity. Those who target infrastructure and efficiency solutions that address the scale and urgency of this transformation are best positioned to benefit.<\/p>\n

Data center differences<\/h2>\n

AI workloads in data centers fall into two distinct categories: training and inference. Training large models involves short, intense bursts of computation, requiring ultra high\u2013density hardware and advanced cooling systems. These workloads typically occur in specialized, hyperscale facilities, often sited in remote locations where abundant, low-cost power is available.<\/p>\n

Inference, by contrast, is a continuous and distributed process. While each query is less demanding, the aggregate electricity use can surpass that of training due to the sheer volume of real-world applications. Inference workloads require data centers to deliver reliable, low-latency performance, often necessitating proximity to population centers and robust network redundancy. This locational preference ensures rapid response times for end users and supports a wider range of commercial and consumer applications.<\/p>\n

From an investment perspective, inference-oriented data centers offer broader applicability and more predictable demand profiles, while training centers\u2014though critical for AI advancement\u2014are more speculative and concentrated in fewer, remote sites. Understanding these distinctions is essential for evaluating infrastructure needs, strategies for sites, and long-term value creation in the evolving AI data center landscape.<\/p>\n

Supply and infrastructure constraints<\/h2>\n

The surge in electricity demand driven by AI and digitalization is colliding with significant supply-side and infrastructure challenges. These constraints are not only slowing the pace of new capacity additions but also creating critical investment opportunities for solutions that can unlock growth and reliability.<\/p>\n

Aging grid infrastructure and grid modernization needs<\/h3>\n

In advanced economies, more than 40% of transmission and distribution assets are more than 20 years old and were designed for centralized, fossil fuel\u2013based generation (Figure 2). The shift toward distributed and intermittent renewables requires a more flexible and intelligent grid. As infrastructure modernizes and digitizes, the risk of cyberattacks and operational disruptions rises, making robust cybersecurity frameworks and resilience planning essential for long-term asset value and risk mitigation.<\/p>\n

\"Stacked<\/p>\n

Supply chain strains: Equipment, labor, and permitting<\/h3>\n

Lead times for essential grid components\u2014transformers, switchgear, and high-voltage cables\u2014have extended from months to years. In the United States, transformer delivery times routinely exceed two to four years, with similar delays in Europe and Asia-Pacific. Transformer costs have doubled since 2020, and the backlog for new orders is at a record high. These delays are driven by surging demand, limited manufacturing capacity, and persistent labor shortages, especially in electrical engineering and construction trades. Tariffs and inflation in materials and logistics are also impacting project economics and timelines, requiring flexible procurement strategies and risk-sharing mechanisms in contracts.<\/p>\n

Permitting and regulatory hurdles compound these challenges. In many jurisdictions, transmission projects face multi-year approval processes. In the United States, the average transmission line takes seven to ten years from planning to completion\u2014a timeline fundamentally misaligned with the rapid buildout of data centers, which can be completed in under two years. This lag is creating acute supply bottlenecks, with some projects forced to rely on local generation or on-site solutions (i.e., battery storage or microgrids) to bridge the gap. Grid-enhancing technologies\u2014such as dynamic line rating and advanced power flow control\u2014can increase the capacity and flexibility of existing infrastructure, offering attractive risk-adjusted returns and faster deployment.<\/p>\n

Figure 3 illustrates the dramatic increase in generation and storage projects waiting in interconnection queues, as well as the average time projects spend in these queues before coming online. This bottleneck is a key factor limiting the speed at which new capacity can be added to the grid. In some jurisdictions, average time in queue is now seven to ten years (e.g., Northern Virginia, Germany, the Netherlands), and in Dublin, new activity is expected to be paused until 2028. Even projects ready to proceed can face multi-year delays, amplifying the risk of supply shortfalls and stranded capital.<\/p>\n

\"Top<\/p>\n

Grid congestion and reliability risks<\/h3>\n

Utilities and grid operators warn that rapid load growth and aging infrastructure could trigger supply shortfalls, price spikes, and curtailments, especially where transmission upgrades lag demand. In major US markets like California, Northern Virginia, and Texas, data center clusters are straining local grids, causing connection delays and even moratoriums. The North American Electric Reliability Corporation (NERC) has flagged data center expansion as a near-term risk to reserve margins. Loudoun County, Virginia\u2014the world\u2019s largest data center hub\u2014is experiencing grid and land constraints, escalating prices, and growing community opposition.<\/p>\n

In Europe, grid saturation in Amsterdam, Dublin, Frankfurt, London, and Paris is limiting the availability of reliable power to new facilities and prompting temporary moratoriums on new data centers. Asia-Pacific faces similar issues, with rapid urbanization and uneven renewable supply causing grid instability.<\/p>\n

Navigating the bottlenecks<\/h3>\n

Supply chain and grid constraints are central for investors, raising the risk of project delays, cost overruns, and stranded assets\u2014especially in congested or slow-permitting markets. These bottlenecks also create opportunities for investment in enabling infrastructure and advanced grid solutions.<\/p>\n

Investors should prioritize regions and projects with clear pathways to grid interconnection, strong utility partnerships, and proactive risk mitigation strategies. As the AI revolution accelerates, the winners will be those who can navigate the bottlenecks\u2014deploying capital not just to meet demand, but to unlock the infrastructure and supply chain solutions that make growth possible.<\/p>\n

Grid and generation solutions<\/h2>\n

Before investors can identify the most attractive opportunities in the evolving electricity landscape, it is essential to understand the strengths, limitations, and deployment realities of major generation and grid technologies. Each solution offers a distinct profile in terms of cost, reliability, scalability, and environmental impact. The optimal mix will depend on local market conditions, regulatory frameworks, and the specific needs of end users. The following overview provides a comparative assessment of key technologies, highlighting factors that drive investment value and the challenges to delivering reliable, low-carbon power at scale.<\/p>\n

Generation technologies<\/h3>\n

Natural gas<\/strong>
\n<\/strong>Natural gas remains the most flexible and scalable dispatchable resource in many markets, offering rapid ramping and high reliability for peak demand and grid stability. However, supply chain bottlenecks (e.g., gas turbines), carbon intensity, and exposure to fuel price volatility present risks, especially as decarbonization policies tighten. Investors should weigh the near-term reliability benefits against the potential for stranded asset risk as emissions standards evolve.<\/p>\n

Renewables (wind and solar)
\n<\/strong>Wind and solar are now the lowest-cost sources of new generation in many regions, but their intermittency creates challenges for grid reliability and requires complementary investments in storage, flexible generation, and advanced grid management. Siting and permitting can be significant hurdles, particularly for large-scale projects. AI can help streamline solar permitting and project management, while evolving battery technologies (e.g., ion-air, flow) offer potential medium-term storage solutions. Investors should consider both the cost advantage and integration challenges.<\/p>\n

In the United States, scaled-back tax credits may depress long-term development, while near-term safe harbor requirements could accelerate construction and create supply shortages and cost increases (Figure 4). Developers with healthier balance sheets will more effectively bring forward project timelines, while liquidity-constrained developers will be more challenged, potentially igniting a buyers\u2019 market for high-quality assets at distressed prices.<\/p>\n

 <\/p>\n

\"Column
\nBattery storage
\n<\/strong>Grid-scale battery storage is rapidly gaining traction for balancing variable renewables and providing ancillary services. Costs have declined and as previously mentioned, battery technologies are improving but are not yet a full substitute for long-duration or seasonal storage. Supply chain constraints and raw material sourcing remain key risks. The investment case is strongest where storage can capture multiple revenue streams or where grid congestion is acute.<\/p>\n

Nuclear<\/strong>
\nNuclear power offers zero-carbon, baseload generation with high reliability. While new large-scale projects face high capital costs, long lead times, and regulatory complexity, interest in small modular reactors (SMRs) is growing (Figure 5). However, a challenge for investors is the proliferation of SMR technologies\u2014more than 80 different designs across four main models\u2014which makes it difficult to achieve standardization and bring down cost curves. Commercial deployment of SMRs remains several years away, and the path to cost competitiveness is uncertain.<\/p>\n

\"Cluster<\/p>\n

Extending the lifetime of existing nuclear plants is already being done in several markets and can be a cost-effective way to maintain reliable, low-carbon power. These life extensions typically require less capital and regulatory risk than new builds and can deliver attractive returns for investors focused on operational upgrades and maintenance. Meanwhile, fusion nuclear energy is attracting considerable private investment but remains in the research phase.<\/p>\n

Geothermal (Next-Generation)
\n<\/strong>Geothermal energy is evolving rapidly, with enhanced geothermal systems (EGS) and closed-loop geothermal expanding the addressable market. EGS leverages oil & gas technologies to unlock geothermal potential in new geographies. Closed-loop systems reuse water and eliminate the need for a continuous supply. Both approaches offer very low-carbon energy and 24\/7 output, with a small land footprint attractive for data centers. Despite high upfront capital requirements, resource risk, extensive transmission infrastructure needs, and long permitting timelines, next-generation geothermal could drive costs lower and enable deployment in new regions. Investors should weigh the long-term potential against these barriers, focusing on projects and platforms that can mitigate development risk and capitalize on technological advances.<\/p>\n

Hydropower
\n<\/strong>Hydropower remains a significant source of reliable, low-carbon electricity in many regions. While new large-scale projects are constrained by geography and environmental concerns, upgrades to existing facilities and pumped storage continue to play an important role in grid flexibility and decarbonization strategies.<\/p>\n

Hybrid systems<\/strong>
\nIncreasingly, developers and utilities are deploying hybrid systems that combine multiple technologies\u2014such as solar and battery storage, wind and gas, or renewables with on-site generation\u2014to deliver firm, flexible power. These integrated solutions help address intermittency, improve grid reliability, and optimize project economics.<\/p>\n

On-site and distributed solutions
\n<\/strong>To address grid constraints and reliability risks, many large power users are investing in on-site generation\u2014such as gas turbines, fuel cells, microgrids, and behind-the-meter storage. These distributed solutions can accelerate project timelines, enhance resilience, and reduce dependence on the broader grid. While they may come at a premium cost and require specialized operational expertise, they are increasingly attractive for data centers and critical infrastructure seeking 24\/7 reliability.<\/p>\n

Grid and data center efficiency innovations<\/h3>\n

Efficiency is a key lever for addressing supply constraints and decarbonization goals. Innovations in grid management and data center operations are unlocking significant value:<\/p>\n