The AI-Driven Energy Transition: Utilities, SMRs, and the New Baseload Revolution

The AI revolution is reshaping global energy demand, creating a seismic shift in how power is generated, distributed, and consumed. As artificial intelligence (AI) data centers expand at breakneck speed, their insatiable appetite for electricity is forcing a reevaluation of energy infrastructure. This transition is not merely a technological evolution but a structural reconfiguration of the energy sector, with unregulated utilities and small modular reactor (SMR) developers emerging as pivotal players. For investors, the stakes are clear: those who align with the infrastructure needs of AI will reap outsized rewards in the coming decade.

The Surge in Data Center Energy Demand

AI data centers are now the largest single driver of electricity consumption growth. In the U.S., data centers accounted for 4.4% of total electricity use in 2024, a figure projected to double or triple by 2028 as AI adoption accelerates. Globally, their energy demand has surged to 415 terawatt-hours (TWh) in 2024, representing 1.5% of global electricity consumption, with projections indicating a doubling to 945 TWh by 2030. This growth is fueled by the computational intensity of AI models, which require accelerated servers consuming 30% more electricity annually.

The U.S. is at the epicenter of this shift. Data center power demand is expected to rise from 35 gigawatts (GW) in 2024 to 78 GW by 2035, while Deloitte estimates AI-driven demand could grow thirtyfold to 123 GW by 2035. These figures underscore a critical challenge: traditional energy systems are ill-equipped to meet the concentrated, high-intensity power needs of AI infrastructure.

Infrastructure Investment: A $1 Trillion Opportunity

The scale of investment required to meet this demand is staggering. Global spending on data center infrastructure is projected to exceed $1 trillion by 2030, with $290 billion already spent in the first half of 2025 alone. Hyperscalers like Microsoft, Amazon, and Alphabet are leading the charge. Microsoft, for instance, has allocated $35 billion to data centers and AI initiatives in 2025, representing 45% of its revenue. These investments are not just corporate expenditures-they are macroeconomic drivers. Harvard economist Jason Furman notes that data center and high-tech investments accounted for nearly all U.S. GDP growth in H1 2025.

Beyond the U.S., the global AI infrastructure boom is expected to require $6.7 trillion in capital expenditures by 2030, with $5.2 trillion dedicated to AI-specific projects. This spending is distributed across IT equipment, facility infrastructure (cooling, electrical systems), and energy solutions. The urgency is compounded by the fact that AI data centers require energy densities rising from 162 kW per square foot in 2024 to 176 kW by 2027.

Unregulated Utilities: NRG and Vistra's Strategic Realignments

As AI-driven demand strains traditional grids, unregulated utilities like NRG and Vistra are repositioning themselves to capitalize on the transition. Vistra has integrated 4.5 GW of nuclear generation into its fleet and secured long-term supply agreements with Amazon and Microsoft. Its focus on baseload generation aligns with AI's 24/7 power needs, and it has hedged 95% of its expected generation for 2025 and 2026.

NRG, meanwhile, is leveraging its acquisition of LS Power to offer flexible generation and virtual power plant infrastructure. This strategy addresses the volatility of AI-driven load fluctuations, particularly in markets like ERCOT. Both utilities are expanding their nuclear and gas-fired capacities to meet the surging demand, with Vistra's 4.5 GW of nuclear generation and NRG's 13 GW of natural gas assets forming the backbone of their AI-focused portfolios.

The SMR Race: NuScale, Terrestrial Energy, and Amazon's Nuclear Ambitions

Small modular reactors (SMRs) are emerging as the linchpin of the AI-driven energy transition. NuScale Power is advancing projects under the U.S.-Japan Framework Agreement, which could unlock $25 billion in investment for advanced nuclear infrastructure. Its collaboration with ENTRA1 Energy underscores the sector's potential to decarbonize energy-intensive industries while meeting AI's power demands.

Terrestrial Energy, which went public via a SPAC in 2025, is also gaining traction. Its Integral Molten Salt Reactor (IMSR) project, developed in partnership with Texas A&M University, offers high-temperature industrial heat and electricity-a dual-use solution ideal for AI infrastructure. The company's $280 million SPAC funding and strategic siting agreements highlight its investment appeal.

Amazon, however, is the most aggressive player in the SMR space. The company is spearheading the Cascade Advanced Energy Facility in Washington state, a project that will deploy X-energy's Xe-100 reactors to generate 960 MW of carbon-free power. Amazon's $1 billion investment in nuclear energy in 2025 includes partnerships with X-energy, Korea Hydro & Nuclear Power (KHNP), and Doosan Enerbility, aiming to mobilize $50 billion in public and private funding. Its long-term goal of securing 5 GW of nuclear capacity by 2039 positions it as a key enabler of the AI energy transition.

High-Conviction Investment Plays

For investors, the AI-driven energy transition presents three high-conviction opportunities:
1. Unregulated Utilities: Vistra and NRG are uniquely positioned to monetize AI's power demands through their diversified generation portfolios and long-term contracts with hyperscalers.
2. SMR Developers: NuScale and Terrestrial Energy offer exposure to the next-generation nuclear infrastructure, with NuScale's $25 billion U.S.-Japan project and Terrestrial's IMSR technology leading the charge.
3. Big Tech's Nuclear Partnerships: Amazon's $1 billion investment in SMRs and its global consortium with KHNP and Doosan represent a strategic bet on nuclear as the backbone of AI infrastructure.

The urgency of this transition is underscored by Morgan Stanley's projection of $2.2 trillion in nuclear investment through 2050-a 40% increase from 2024 forecasts. With AI-driven data centers projected to require tenfold more power over the next five years, the window to invest in scalable, resilient energy infrastructure is narrowing.

Conclusion

The AI-driven energy transition is not a distant future-it is here, reshaping grids, utilities, and investment landscapes. For those who recognize the scale of this shift, the path forward is clear: invest in infrastructure that can meet the relentless demands of AI while ensuring sustainability and resilience. Unregulated utilities, SMR developers, and Big Tech's nuclear partnerships are the cornerstones of this new energy era. As the world races to power the AI revolution, the winners will be those who build the infrastructure to fuel it.