Small modular reactors (SMRs) have become the nuclear industry's most credible answer to reversing three decades of cost overruns and delays that made large reactors economically unviable in Western markets. By shifting construction from massive on-site megaprojects to factory-fabricated modular units, SMRs promise to cut capital costs from $8,000-12,000 per kilowatt for recent large reactor builds down to $4,000-7,000 per kilowatt, with construction timelines compressed from 6-10 years to just 3-5 years. But the real story isn't about engineering breakthroughs; it's about desperation. Tech giants like Microsoft, Amazon Web Services, and Google need massive amounts of uninterrupted, 24/7 power that solar and wind alone cannot reliably provide without exorbitant battery costs. That urgency is reshaping the entire nuclear industry.

What Happened to the First Generation of Western SMR Projects?

The cautionary tale begins with NuScale's Carbon Free Power Project in Idaho, the only SMR design to achieve U.S. Nuclear Regulatory Commission (NRC) design certification. The project was terminated in November 2023 after its target electricity price ballooned from $58 per megawatt-hour to $89 per megawatt-hour. That's a 53% cost overrun that institutional investors cannot ignore. The lesson is stark: first-of-a-kind (FOAK) SMR costs will substantially exceed the industry's optimistic targets. Experts project first-generation units will cost $90-120 per megawatt-hour, compared to the long-term goal of $60-80 per megawatt-hour once factories reach their tenth unit and achieve economies of scale.

This gap between promise and reality has forced vendors to pivot their entire business strategy. Instead of selling to traditional utility companies, NuScale, Oklo, and other SMR makers are now targeting private data center operators who possess the capital to underwrite the first-of-a-kind premium and can negotiate long-term power purchase agreements directly with reactor operators. AWS's acquisition of a data center campus powered directly by the Susquehanna nuclear plant signaled this paradigm shift; Big Tech is willing to pay a premium for reliable, behind-the-meter nuclear power.

Which SMR Designs Are Actually Moving Forward?

Three Western designs are in advanced licensing or deployment stages, and all are light-water reactors (LWRs), avoiding the regulatory uncertainty of more exotic designs. Here's where each project stands:

• GE Hitachi BWRX-300 at Darlington, Ontario: This 300-megawatt boiling water reactor is the closest to construction in North America. Ontario Power Generation submitted a construction license application, with a target deployment date of 2029.
• Rolls-Royce SMR in the United Kingdom: A 470-megawatt pressurized water reactor design undergoing Generic Design Assessment (GDA), with target completion in 2026 and deployment in the early 2030s.
• NuScale VOYGR: Achieved NRC design certification in January 2023 but is now seeking new offtakers after the Idaho project's collapse. The design can scale from 77 megawatts per module up to 924 megawatts with 12 modules.

Meanwhile, non-Western markets are moving faster. Russia already operates the Akademik Lomonosov, a 70-megawatt floating SMR powering the Arctic region, and China's CNNC Linglong One (ACP100), a 125-megawatt land-based SMR in Hainan, is structurally complete and on track to become the world's first dedicated commercial land-based SMR. This divergence reveals a stark reality: state-backed financing and streamlined regulatory environments in China and Russia are outpacing Western market-driven frameworks.

How Are Tech Giants Financing This Nuclear Pivot?

U.S. policy is quietly reshaping the economics. The Inflation Reduction Act (IRA) provides a production tax credit of $15 per megawatt-hour for existing nuclear plants and a technology-neutral clean electricity credit for new zero-carbon generation, effective in 2025. For an SMR generating electricity at $60-80 per megawatt-hour, that $15-25 per megawatt-hour credit represents a 20-40% effective subsidy, materially improving SMR competitiveness in U.S. markets. The European Union has also moved nuclear into its taxonomy of sustainable activities, enabling access to green bond financing for SMR projects in EU member states.

Microsoft's aggressive hiring of nuclear experts and its backing of TerraPower, the advanced reactor company founded by Bill Gates, demonstrates a long-term strategy to integrate SMRs directly into data center campus designs. This isn't a short-term energy fix; it's a fundamental bet that nuclear will become as essential to AI infrastructure as GPUs.

What's the Realistic Timeline for Global SMR Deployment?

The International Energy Agency (IEA) projects 10-25 gigawatts of installed SMR capacity globally by 2035, representing just 1-3% of global nuclear capacity. That's a steady evolution, not an overnight revolution. The first Western commercial SMR is targeted for 2029 at Darlington, but regulatory timelines vary significantly. Light-water reactor designs typically require 6-8 years for NRC certification, while non-light-water designs face 8-12 years or more. Advanced Generation IV reactors like molten salt and sodium-cooled designs promise greater efficiency and industrial heat applications, but they face critical supply chain bottlenecks, particularly for High-Assay Low-Enriched Uranium (HALEU) fuel, pushing their commercial deployment into the late 2030s.

How to Evaluate SMR Projects as an Investor or Policymaker

• Distinguish FOAK from NOAK costs: First-of-a-kind SMR projects will cost 30-50% above the industry's long-term targets. Expect $90-120 per megawatt-hour for early units, not the promised $60-80.
• Assess regulatory maturity: Light-water reactor designs have clearer regulatory pathways (6-8 years for NRC certification) compared to advanced reactors, which face 8-12+ years of review.
• Monitor fuel supply chains: Advanced Generation IV reactors depend on specialized fuels like HALEU, which face production bottlenecks that could delay deployment by years.
• Track data center offtake agreements: The most viable near-term SMR projects are those with long-term power purchase agreements from tech giants, not traditional utilities.

The nuclear-AI convergence is real, but it's not a silver bullet. SMRs represent the industry's most credible pathway to reversing decades of cost escalation, yet the first wave of Western projects reveals a sobering truth: the gap between engineering ambition and economic reality remains substantial. For AI data centers desperate for reliable, carbon-free power, that premium may be worth paying. For traditional utilities and policymakers betting on rapid, cost-effective decarbonization, the timeline and expense demand realistic expectations.