Nuclear New-Build and SMRs: The Investment Case

Introduction

Nuclear energy is experiencing a revival that few in the energy industry anticipated five years ago. The convergence of two forces, the explosive growth in electricity demand from AI data centers and the need for 24/7 carbon-free generation, has transformed nuclear from a sector in managed decline to one of the most active areas of energy transaction activity. For energy investment bankers, nuclear is generating deal flow across multiple categories: plant restart transactions, long-term corporate PPAs with technology companies, SMR development financing, and M&A involving nuclear fleet operators like Constellation Energy. Understanding the economics, the challenges, and the deal dynamics of nuclear is increasingly essential for anyone covering power and utilities.

The scale of the opportunity is significant. Big tech companies have signed over 10 GW of new US nuclear capacity commitments in the past year. The IEA expects US data center power consumption to increase by 130% from 2024 to 2030, and nuclear energy could meet up to 10% of data center electricity demand by 2035. This demand is not theoretical: it is backed by multi-billion-dollar corporate commitments from Microsoft, Google, Amazon, and Meta, each of which requires reliable, baseload, carbon-free power that solar and wind alone cannot provide on a 24/7 basis without massive storage investment.

The near-term investment opportunity in nuclear centers on the existing fleet of operating reactors and potential plant restarts, not new construction. The US has 93 operating commercial reactors generating approximately 775 TWh per year, roughly 19% of total US electricity. These plants were built decades ago at high capital cost, but their operating costs are low ($25-35/MWh all-in, including fuel, maintenance, and regulatory costs), making them highly competitive generation assets in current markets.

The most significant recent transaction is Microsoft's 20-year, $16 billion power purchase agreement with Constellation Energy to restart the 835 MW Unit 1 at the Three Mile Island site in Pennsylvania (renamed the Crane Clean Energy Center). The facility is targeting a 2027-2028 restart. This deal is remarkable for several reasons: it represents the first restart of a permanently shut-down US nuclear plant, it demonstrates that a technology company is willing to commit $16 billion over 20 years for carbon-free baseload power, and it created significant market value for Constellation's stock (which more than doubled following the announcement and related nuclear optimism).

Other major nuclear-tech deals include Meta's 20-year PPA with Constellation for 1.1 GW from the Clinton Clean Energy Center in Illinois, Google's agreement with Kairos Power for approximately 500 MW of SMR capacity across six to seven reactors (deployment beginning after 2030), and Amazon's $20 billion+ investment in nuclear-powered data center infrastructure, including a deal to purchase 1.9 GW through 2042 from Talen Energy's nuclear plant in Pennsylvania.

SMRs are the technology that nuclear proponents believe can overcome the cost and timeline challenges that have plagued conventional nuclear construction. The concept is to build smaller reactors (77-300 MW per module) in factories, transport them to sites, and assemble them, reducing the custom engineering, on-site construction labor, and schedule risk that drove Vogtle (the most recent US conventional nuclear project) to over $35 billion and more than seven years of delays.

Small Modular Reactor (SMR)

A nuclear reactor with an electrical output of less than 300 MW per module, designed for factory fabrication and modular on-site assembly. SMRs use various reactor technologies (pressurized water, molten salt, high-temperature gas, sodium-cooled) and are intended to be deployed in configurations of multiple modules at a single site, allowing capacity to scale incrementally. The modular approach is expected to reduce per-unit capital costs through manufacturing learning curves, standardized designs, and shorter construction timelines. NuScale Power's 77 MW pressurized water reactor module is the only SMR design to have received full design certification from the US Nuclear Regulatory Commission (NRC).

NuScale Power, the most advanced Western SMR developer, received NRC design certification for its 77 MW module. However, the cost trajectory has been concerning. NuScale's estimated construction cost for its first project (the canceled Carbon Free Power Project in Idaho) rose from $5.3 billion to $9.3 billion, translating to approximately $20,000/kW. This is comparable to the final cost of the Vogtle conventional nuclear project and undermines the central argument that SMRs would be cheaper per kilowatt than large reactors. The target power price rose from $58/MWh to $89/MWh (and would be significantly higher without approximately $4 billion in federal subsidies, including a $1.4 billion DOE contribution and a $30/MWh IRA production tax credit).

Romania's Doicesti project represents the first committed NuScale SMR deployment: a 462 MW facility using six NuScale modules at a former coal plant site, with Nuclearelectrica's shareholders approving the final investment decision. In the US, the Department of Energy has committed approximately $800-900 million to support SMR development, and Google's Kairos Power deal represents the first major US corporate commitment to SMR technology.

The investment community remains cautiously optimistic but cognizant of execution risk. No SMR has been built at commercial scale in the Western world (Russia's floating Akademik Lomonosov and China's HTR-PM are the only operating SMRs globally). The manufacturing learning curve that SMR proponents cite as the key cost reduction mechanism has not yet been demonstrated. Capital costs will need to decline substantially from current estimates before SMRs can compete with solar-plus-storage or natural gas generation on pure economics, though the 24/7 carbon-free attribute provides value that intermittent renewables cannot match.

Nuclear assets are valued differently from other generation technologies. Operating nuclear plants with remaining useful life of 20-40 years (following license extensions) are valued as long-duration contracted or quasi-contracted cash flow streams. In regulated markets, nuclear plants earn their allowed return through the utility's rate base. In merchant markets, nuclear plants sell into wholesale electricity markets and earn additional revenue from capacity markets, zero-emission credits (ZECs), and, increasingly, long-term corporate PPAs.

The premium that data center operators are willing to pay for 24/7 carbon-free power is what makes nuclear economics work. Microsoft's willingness to commit $16 billion over 20 years for a single 835 MW nuclear plant implies a PPA price well above the plant's marginal operating cost, reflecting the scarcity value of reliable, carbon-free baseload generation in a grid increasingly dominated by intermittent renewables.

For energy bankers, these risks create both challenges and opportunities. The risk complexity of nuclear transactions creates advisory fees that exceed those for simpler renewable energy deals. The need for government support means that nuclear project finance involves DOE loan programs, state-level regulatory approvals, and multi-stakeholder negotiation processes that require specialized advisory capabilities.