That era is ending. 

AI is one visible driver of new electricity demand, but it is far from the only one. We are electrifying more of the economy. Manufacturing is returning to the United States. Robots and physical AI are moving quickly from research labs into the real world. 

We are simply going to need far more electricity than we have today. And that is why we need to re-evaluate nuclear energy.

At NVIDIA, we see both sides of this transition. The AI and accelerated computing we develop with our broad ecosystem of partners can also help energy developers design, simulate and operate infrastructure. This perspective that shapes our belief that the nuclear debate must move beyond whether to build—and focus on how technology can help the industry build safely, faster and at scale.

Why Nuclear, Why Now 

Support for nuclear power is converging in Washington in a way we have not seen in decades. Nuclear energy is increasingly viewed as a tool for mitigating climate change, while meeting increasing demand for energy abundance and economic growth.

But the case for domestic nuclear expansion is stronger.

Nuclear promises clean, firm, abundant energy that can meet growing demand and serve large industrial loads. The nation that sets the standard for safe, affordable, modern nuclear power will have strategic, security and commercial advantages.

France’s long-term commitment to nuclear has already demonstrated that it can provide safe and reliable power at scale and maintain public support. AI is now accelerating interest by creating a massive, well-capitalized customer base that is willing to invest in clean power and next-generation nuclear technologies.

A New Wave of Investment and Innovation

The funding behind this new era of nuclear is coming from a mix of federal agencies and national laboratories, venture-backed startups, and companies investing in the clean power they will need. The first megawatt of a new nuclear technology is extraordinarily expensive, but once a design is developed, tested, and proven viable, economies of scale can help bring costs down.

Companies are pursuing everything from advanced fission technologies like microreactors and small modular reactors to fusion. At NVIDIA, we’re witnessing far more demand for energy than available supply, in addition to construction and timing delays. Designing reactors, interpreting technical and regulatory information, simulating plant operations and integrating new power sources all require the ability to process increasingly complex data. NVIDIA works across that entire computing stack, giving us a view of how these pieces can work together.

TerraPower, for example, is connecting an NVIDIA Omniverse-powered platform to use digital twins to compress parts of advanced nuclear siting and design. Atomic Canyon is using accelerated computing to build AI-powered knowledge tools for reactor operators, turning procedures, regulatory guidance, licensing records, and other nuclear data into a usable knowledge layer. Commonwealth Fusion Systems is using Omniverse libraries and OpenUSD to compress years of fusion experimentation into weeks. 

We do not yet know which approaches will prove most successful. We need an environment where they can safely compete and prove what works — while rebuilding U.S. industrial capacity, skilled talent, and manufacturing supply chains along the way.

A New Technological Era Requires a New Risk Calculus 

Safety remains the central hurdle. But we are in a new technological era of reactor design, and the old risk calculus no longer reflects what many of these systems are designed to do.

Many advanced reactors are being built with passive safety mechanisms, where the fail-safe state is the default rather than something dependent on active intervention.

Waste management is also part of the equation, with advances in reactor design and waste management helping minimize radioactive waste and safely contain what remains.

The right approach is to minimize both the probability and magnitude of an incident: make failures extremely unlikely while ensuring even worst-case scenarios can be contained and mitigated. Modular and advanced designs could also enable smaller, safer facilities with broader applications.

AI can play a role in that work, but only within clear boundaries. Argonne National Laboratory is testing PRO-AID, a system that combines AI with physics-based models to identify and explain equipment problems in real time. In testing, it detected simulated sensor and component failures as they occurred, while leaving verification, response and operational decisions to trained personnel under established safety and regulatory procedures.

The government also has a tremendous opportunity to build public trust. Early projects and pilots at Idaho National Laboratory and Department of Defense sites demonstrate that advanced nuclear technologies can operate safely and reliably in the real world. 

How Do We Accelerate the Deployment of Nuclear Energy? 

The biggest constraint on nuclear development today is regulation. Many of the rules governing nuclear were built around older reactor designs and do not reflect the technical reality of the advanced systems being developed today. 

There are signs of progress. The NRC has moved toward 18-month review milestones for new reactor licenses and construction permits. TerraPower’s Kemmerer 1 construction permit, originally scheduled for a 27-month review, received its final safety evaluation in 18 months.

Now we need to make that speed and predictability the norm. The NRC should continue moving toward risk-informed regulation with predictable review timelines appropriate to advanced technologies. Congress should ensure the bipartisan ADVANCE Act, signed into law in July 2024, is fully implemented by providing the resources, oversight, and direction needed to make its reforms real. That includes completing work on efficient license reviews, creating an expedited process for qualifying new reactor applications, finalizing lower fees for advanced-reactor applicants, and developing clear frameworks for microreactors, fusion machines, advanced fuels, and environmental reviews. The NRC will also need the technical staff and budget flexibility to do this work effectively. The law provides the direction; the remaining challenge is turning it into rules, guidance and review practices that developers can rely on.

DOE can help technologies prove themselves before commercial deployment. Its Reactor Pilot Program recently enabled Valar Atomics to achieve criticality with an advanced reactor in Utah — the first DOE-authorized reactor built outside a national laboratory. More pathways like this can help new designs move from testing toward commercialization.

We should not lower the bar for nuclear safety. We should make the path to meeting it clearer, faster and more predictable.

Timing Nuclear Against AI Demand

Even with these changes, nuclear will not come online fast enough to meet near-term AI energy needs. Advanced projects take years to develop, approve, and build, making an all-of-the-above approach the realistic near-term answer.

Some nuclear companies are finding ways to move faster through off-the-shelf components, simpler and repeatable designs, and less reliance on traditional nuclear supply chains. Faster licensing, more opportunities to demonstrate new reactors and designs built for repeatability can help turn those individual efforts into an industry capable of scaling. 

Nuclear can then play a much larger role in the medium to long term as technologies mature and costs come down.

The opportunity extends well beyond powering AI. A revitalized nuclear industry could strengthen U.S. manufacturing, create new markets for American technology, and help secure the country’s position in an increasingly energy-intensive global economy.

Nuclear’s contribution to AI will ultimately be measured in clean, reliable power. NVIDIA’s broader clean-energy work, from AI-assisted grid interconnection to flexible battery storage, reflects the supporting infrastructure nuclear will need as it connects to a more complex, energy-intensive grid. Together, these technologies can strengthen U.S. manufacturing and help secure the country’s position in the global energy economy.The U.S. has the capital, technology, and talent to lead. Now we need to create the conditions to build. A nuclear renaissance will take time, which is exactly why we need to start accelerating it today.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.