Nuclear Delays: Not Always Financial in Nature
Let's take a look at the US nuclear industry and regulatory approach (we will probably have a few posts on this topic).
Data centers are the backbone of the modern economy. They power everything from cloud services to AI models, and as demand for AI, machine learning, and big data grows, so does the need for massive amounts of energy. But here's the issue: data centers can’t afford to wait decades for new power plants.
Nuclear energy, particularly small modular reactors (SMRs), can provide a clean, reliable, and scalable energy solution — we have talked about this before. However, the U.S. nuclear licensing process is bogged down by outdated regulations, resulting in long approval timelines that delay the deployment of much-needed power solutions. The question is: Why does it take so long?
The Real Timeline for Nuclear Construction: Vogtle Units 3 & 4
Let’s take a look at Vogtle Units 3 & 4—a real-time case study of nuclear’s painfully slow process. These reactors are a testament to the sluggish nature of nuclear power deployment in the U.S. They’re also a reminder of why data centers are worried about nuclear being a viable partner in this quest for energy.
Here’s the actual timeline for the Vogtle project:
2012: The NRC issued the Combined Operating License (COL) for Vogtle. This means the project got the green light to begin construction. That’s 4 years after Southern Nuclear Operating Company (SNC) filed its application for combined licenses for two AP1000 reactors here, in Georgia. Just for reference, Part 52 of the licensing process took years to navigate before construction could even begin.
2013: Construction of Unit 3 began—finally. And by “began,” I mean there were tons of design revisions, procurement delays, supply chain issues, and more inspections than anyone could care to count. In a nutshell: delays. Note: the original construction start date was 2013 with operational reactors by 2016 and 2017 (units 3 and 4). But the construction delays pushed the actual start dates back repeatedly.
2023: Unit 3 came online in March 2023, marking the first new nuclear reactor in the U.S. in 30 years. Yep, three decades. It’s kind of like waiting for the second coming of nuclear energy.
2024: Unit 4 was finally operational in April 2024. That’s 13 years after initial approval. Talk about taking your time. The entire process from approval to operation spanned over a decade—and that’s with all the technical issues and cost overruns.
Why This Matters for Data Centers (and AI)
Now, here’s where the rubber meets the road: Data centers don’t have 13-17 (depending how you look at the process) years to wait for a power source. With AI-driven infrastructure demanding more compute power every day, energy consumption is projected to grow exponentially. By 2030, data centers are expected to consume 3-4% of global electricity. That's a problem when we can’t even get nuclear power online in under a decade.
And here’s why nuclear energy, particularly small modular reactors (SMRs), is the ideal solution for this data center energy crisis:
SMRs are compact, efficient, and inherently safer than traditional reactors.
SMRs offer clean, scalable energy to meet the growing demand.
SMRs can be deployed closer to energy-hungry facilities, cutting down on transmission losses.
But, in order to get these reactors up and running, the licensing process needs a massive overhaul.
The Licensing Bottleneck: How Long Is Too Long?
The current U.S. nuclear licensing system is, frankly, a mess. It’s bloated, bureaucratic, and overly complicated. Vogtle’s timeline is a perfect example of how traditional licensing frameworks slow progress. The process takes years—even decades—because we’re still operating under Part 50 and Part 52, designed for large traditional reactors, not modular, advanced designs.
Take a look at NuScale, the first SMR design to undergo licensing in the U.S. It took over 10 years for approval, despite being designed for faster deployment. This highlights the core issue: our nuclear licensing process isn’t designed to keep up with modern needs. And stay tuned for the next post, where we will talk about SMRs taking different regulatory paths for licensing.
We need Part 53—a risk-based, performance-driven framework that will speed up approvals for advanced reactors and allow designs like SMRs to get to the construction phase faster. But that’s still a work in progress. Part 53 has not been fully implemented yet, and it remains stuck in regulatory purgatory.
What’s Holding Us Back? The Regulatory Quagmire
Let’s talk about what’s keeping us from getting to faster nuclear power:
Part 50: The traditional approach, requiring site-specific reviews, environmental assessments, and exhaustive safety protocols—all necessary but incredibly time-consuming. Vogtle’s 13-year timeline is a direct result of this process.
Part 52: This was supposed to speed things up by allowing pre-approved reactor designs. But it still requires site-specific reviews and redundant safety checks. Even with NuScale’s SMR, the process still took 10 years.
NEPA (National Environmental Policy Act): The requirement for environmental assessments at each stage of reactor licensing adds unnecessary delays and can be used as a loophole for anti-nuclear opponents to slow down projects.
NRC's Slow Response: The NRC itself has long review times, which means reactor developers have to wait months or years just for one safety review to come through.
The whole thing is like an infinite loop of red tape that can only be broken by reform.
What Can Be Done?
Here’s the kicker: we don’t need to reinvent the wheel. We need to look at what other countries are doing and borrow their best practices:
Russia and China are building reactors at lightning speed because they’ve streamlined their processes, standardized reactor designs, and embrace batch approvals for new reactors - more on that is coming in this substack.
Canada has shown us that we can be agile in licensing new designs with their modular reactor framework.
It’s time to cut the fat and get nuclear power up and running faster. SMRs are the key to clean energy, but we need a new regulatory framework to make that happen. Part 53 could be the answer, but it needs to be implemented fasterand more effectively.
No More Waiting—Data Centers Need Energy Now
Data centers and AI infrastructure can’t wait decades for nuclear power. They need clean, scalable energy solutions now. The nuclear licensing process is broken and needs to be fixed—fast.
In the next post, we’ll dive into Part 53, what it offers, and why it’s critical for the future of U.S. nuclear energy. We will also explore how SMRs have tried both 10CFR50 and 52—with different results. Stay tuned.
Join the Conversation
How do you think the U.S. nuclear licensing process should be reformed? Share your thoughts below, or pass this alongto anyone who is interested in speeding up nuclear deployment for data centers and the future of energy.
The nuclear build timeline is likely the quest for build capabilities which links to both financial and supply chain. Data centers need energy today and making decisions today. Nuclear vendors must be very smart to stay relevant.