SMR Data Center Projects, 11 Confirmed 1 GW Sites, $900 M DOE Fund, and 3 Big Tech Nuclear Deals (2024 to 2026)

Grid Bypass Strategy, SMR and Geothermal Data Center Projects

The explosive power demand from artificial intelligence is rendering the traditional model of grid-dependent data centers obsolete, forcing a strategic migration toward integrated campuses where gigawatt-scale computation is co-located with dedicated nuclear and geothermal power generation. This shift is not a choice but a necessity, driven by saturated transmission corridors and multi-year interconnection queues that make reliance on the public grid a primary inhibitor to growth. Projects like the Creekstone Energy and Blu Sky AI campus in Utah are the blueprint for this new paradigm: bypassing the grid entirely to secure the vast, reliable, and carbon-free power required for AI infrastructure.

• Between 2021 and 2024, the core problem intensified as hyperscalers recognized that grid limitations were a direct threat to their expansion plans. Grid connection backlogs grew by 30% in 2023 alone, with wait times in key markets like Northern Virginia extending to 4-8 years. Early responses involved large-scale Power Purchase Agreements (PPAs) with existing nuclear facilities, such as Amazon Web Services’ (AWS) deal with Talen Energy.
• From 2025 to today, the market has pivoted from procuring power from the existing grid to creating new, private energy ecosystems. This is validated by the emergence of at least 11 confirmed data center projects globally with planned capacities of 1 GW or more. Major technology firms are now leading this transition through direct action; for instance, Google is planning multiple Small Modular Reactor (SMR) projects, and Meta has established clear targets for adding gigawatts of new nuclear capacity to power its future operations.
• The development of enhanced geothermal systems (EGS) provides a critical, renewable 24/7 power source to complement nuclear. Commercial validation for this approach is accelerating, with companies like Fervo Energy proving EGS can be developed at gigawatt scale and securing partnerships with hyperscalers like Meta. This dual-technology strategy, combining nuclear and geothermal, represents the most resilient solution for achieving energy independence.

$1.7 T in Capital, Data Center Infrastructure Investment to 2030

The monumental power requirements of AI are catalyzing an equally large investment cycle, with projections indicating over $1.7 trillion in capital expenditures for data center infrastructure by 2030. This capital is increasingly being allocated not just to servers and buildings, but to the underlying energy generation and transmission assets. Federal incentives and private capital are aligning to finance the development of dedicated power sources, transforming energy from a simple operating expense into a core, integrated asset for AI companies.

• The scale of financial commitment is immense, with over $1 trillion in private capital earmarked for building out new AI infrastructure. This funding is essential to cover the high upfront costs of developing power plants, particularly advanced nuclear and geothermal projects.
• Government support is a critical enabler for these capital-intensive projects. The U.S. Department of Energy (DOE) is actively encouraging the development of SMRs through a $900 million funding opportunity aimed specifically at supporting the deployment of next-generation reactors.
• Federal tax incentives under the Inflation Reduction Act, including the Clean Energy Production Tax Credit (§45 Y) and Investment Tax Credit (§48), substantially improve the financial viability of new nuclear and geothermal projects. These credits provide long-term revenue certainty and reduce the initial capital burden, making it easier to secure project financing.

Table: Strategic Investments and Financial Commitments

Big Tech Nuclear Partnerships, 3 GW in PPA and SMR Deals

To circumvent grid constraints and secure stable, carbon-free power, hyperscalers are forming direct, long-term partnerships with nuclear and geothermal energy developers. This model allows them to lock in power at predictable costs and ensure supply for decades, a strategic advantage that is impossible to achieve through the volatile spot electricity market. These alliances signal a fundamental restructuring of energy procurement, where the largest consumers are now driving the development of new generation assets.

• Amazon Web Services (AWS) established a significant precedent by acquiring a data center campus adjacent to Talen Energy’s Susquehanna nuclear plant, with an agreement to procure up to 1, 920 MW of 24/7 nuclear power.
• Google has taken a direct role in developing future nuclear capacity, with plans for three 600 MW nuclear projects to power its data centers and a specific partnership with Kairos Power to deploy SMRs by 2030.
• Meta has publicly stated its intent to facilitate the addition of 1-4 GW of new nuclear capacity in the U.S. starting in the early 2030 s, viewing it as essential for powering future AI innovation.

Table: Data Center Energy Partnerships

Central Utah, The New Gigawatt Data Center Hub

The geographic calculus for data center location has fundamentally inverted, shifting away from power-constrained, high-cost urban markets toward resource-rich regions like Central Utah. The decision to site massive campuses in these areas is driven by the logic of moving computation to the energy source. Access to land, water, and favorable geology for both geothermal and nuclear development makes locations like Utah ideal for building the self-sufficient, gigawatt-scale energy and compute ecosystems that AI demands.

• Between 2021 and 2024, the industry witnessed the consequences of concentrating development in traditional hubs like Northern Virginia. Power availability became a severe constraint, with utilities implementing rationing and halting new connections, forcing developers to look elsewhere.
• From 2025 onward, a clear migration to new territories began. Utah is now at the forefront of this trend, with 2, 602 MW of data center capacity currently under construction, nearly triple its existing operational capacity. This includes the massive Creekstone Delta Gigasite, establishing the region as a critical hub.
• The strategic importance of this geographic shift is underscored by research from institutions like Idaho National Laboratory, which has identified the Intermountain West region as particularly suitable for large-scale, integrated nuclear and renewable energy research and deployment.

SMR and EGS Maturity, Creekstone’s Bet on Next-Gen Power

The viability of gigawatt-scale, privately powered data center campuses hinges on the commercial readiness of two key technologies: Small Modular Reactors (SMRs) and Enhanced Geothermal Systems (EGS). While conventional nuclear and geothermal power are mature, they lack the scalability and siting flexibility required for this new model. The industry’s massive investments in SMRs and EGS represent a calculated decision that these technologies are sufficiently mature to transition from pilot-scale to full commercial deployment within the compressed timelines demanded by AI growth.

• SMR technology has advanced from design concepts to active commercialization efforts. Between 2021-2024, the focus was on regulatory certification for the more than 80 designs in development. Since 2025, major commitments from buyers like Google and direct government funding have shifted the focus to deploying first-of-a-kind (FOAK) commercial units, although execution risks related to construction and supply chains remain a primary concern.
• EGS technology is proving its ability to unlock geothermal potential on a massive scale. Previously an R&D-focused field, EGS is now commercially viable, as demonstrated by companies like Fervo Energy which are successfully developing projects at a scale relevant to data centers. The DOE’s projection that geothermal could supply up to 120 GW in the U.S. by 2050 is heavily dependent on the continued success of EGS.

SWOT Analysis, The Integrated Energy-Compute Model

The integrated energy-compute model pioneered by projects like the Creekstone-Blu Sky campus presents a powerful strategic response to the AI-driven energy crisis, but it is not without substantial risks. Its strengths lie in securing unparalleled operational resilience and cost predictability. However, it faces significant threats from complex regulatory environments and the financial uncertainties associated with deploying nascent, capital-intensive technologies.

• Strengths in energy independence and cost stability are the primary drivers, creating a significant competitive advantage.
• Weaknesses are centered on the high capital expenditure and reliance on first-of-a-kind technologies with unproven construction timelines and costs.
• Opportunities are created by massive federal incentives and the ability to set a new, highly defensible industry standard for AI infrastructure.
• Threats include multi-year regulatory delays, potential for major construction cost overruns, and resource constraints like water rights in arid regions.

Table: SWOT Analysis for the Integrated Energy-Compute Model

Creekstone Project 2027, Phased Geothermal Deployment Scenario

The most credible path forward for the Creekstone campus, and similar ventures, involves a phased deployment where the more technologically mature and faster-to-permit energy source comes online first. This suggests the geothermal component will likely provide the initial power blocks to begin data center operations in the 2027-2029 timeframe, while the larger, multi-gigawatt nuclear expansion follows a longer-term, post-2035 development track dependent on successfully navigating the federal regulatory process.

• If this happens: Watch for a final investment decision (FID) on the geothermal phase of the project within the next 12-18 months. This would be the strongest signal that the phased strategy is in motion.
• Watch this: Key near-term signals include applications for drilling and site preparation permits for the geothermal component in Utah. For the nuclear portion, the first major milestone will be pre-application engagement with the Nuclear Regulatory Commission (NRC), which will reveal the selected SMR technology and the official start of the long licensing journey.
• These could be happening: Competitors, seeing the strategic logic, will accelerate their own land and resource acquisition in other suitable states like Wyoming, Idaho, and Texas. This will trigger a wave of announcements for similar integrated energy-compute campuses, creating a new competitive dynamic centered on securing power, not just building data centers.

The questions your competitors are already asking

This report covers one angle of the shift toward grid-independent, gigawatt-scale data center infrastructure. The questions that matter most depend on your work.

• Which companies are gaining or losing ground in the nuclear-powered data center market?
• What is the outlook for SMR deployment in AI data centers by 2030?
• How does geothermal power compare to nuclear SMRs for powering gigawatt-scale data centers?
• Which hyperscalers are actively adopting the grid bypass strategy with on-site nuclear and geothermal power?

This report does not answer these. Enki Brief Pro does.

Your question, your angle, your framework. SWOT, PESTL, scenario modelling. The same niche depth, built around the decision your work actually depends on.

Run your first brief in Enki Brief Pro