Nuclear: all the rage for big tech

January 2025  |  SPECIAL REPORT: ENERGY & UTILITIES

Financier Worldwide Magazine

January 2025 Issue

Artificial intelligence (AI) is certainly making headlines. But how will all these computations be powered without contributing to climate change? After years of intense competition to develop AI and cloud computing networks, the practical carbon-free energy challenges of offering these products are now catching up to technology businesses.

As always, big tech innovates. A new and exciting opportunity that has drawn the attention of big tech companies is a keen interest in nuclear energy, and they are putting their money on the table. If anyone had predicted three years ago that the mothballed Three Mile Island (TMI) Unit 1 would be brought back to life courtesy of a power purchase agreement with Microsoft, they would have been laughed out of the room. But that is exactly what is happening, with utility Constellation Energy restarting TMI Unit 1 to power data centres.

On reflection, it does in fact make perfect sense that nuclear power is where the world’s tech behemoths are eyeing up their next investment push. After all, in order to offer AI and cloud products to mass markets, you need to have vast amounts of data centre capacity, and these data centres require enormous amounts of power to run. All the while, both consumer and regulatory scrutiny around environmental, social and governance and sustainability has been heating up, and tech companies themselves continue to see value in producing and operating their products in a sustainable way.

Toward the tail end of 2024, we saw technology companies identify nuclear as a potentially ideal way to square the circle of needing ever more electricity. However, is there even further hidden value in having this new base of deep-pocketed potential investor-customers? Moreover, while we are beginning to see cash flowing, is it too soon to say that a tech-investment-fuelled nuclear renaissance is fully underway?

The need for nuclear in tech

As tech companies have continued their push to develop groundbreaking new AI and cloud computing services, the scale of the energy required to enter this new age of consumer technology products has become evident.

Some reports indicate that AI web searches require up to 10 times the amount of energy needed for a standard web search, and the required energy for this end-user service of AI is only scratching the surface. Generative AI and machine learning tools need to be trained, and this can require even more vast amounts of clean power.

As AI continues to proliferate across all corners of business, the power demands of the technology are only set to grow exponentially – an issue which the tech giants pioneering AI development are acutely aware of. Of course, the wider global push to transition to clean energy means that there is a variety of sustainable power solutions available. However, if they look to wind or solar power, tech companies will quickly encounter thorny challenges.

For example, the UK has a fairly high penetration of wind and solar power, meaning the infrastructure is in place and so these solutions could be seen as a potentially ideal starting point for tech companies looking to decarbonise. But even AI cannot control the weather. If the sun is not shining or the wind is not blowing, there is a real risk that these technologies will not provide enough clean energy to keep operations going. Periods when renewables are not available occur often enough that the Germans have invented a word for it: ‘dunkelflaute’. A dunkleflaute translates as something like ‘dark doldrums’ – a period of time when there is little to no energy generation from wind and solar power.

Nuclear plants, on the other hand, produce a reliable and uninterrupted supply of clean energy around the clock; ideal for powering services that need to be in constant operation, such as Google searches or Amazon Web Services.

Options and challenges

When it comes to a viable route into investing into nuclear power for big tech companies, there are more options today than ever before.

One option is to look for existing large reactors that can be taken over through direct investment or whose power output can be purchased in whole or in part through power purchase agreements (PPAs), and to then set data centre operations nearby so that they are powered directly by the nuclear plant. This is what is spurring the restart of TMI Unit 1.

A second way would be to use PPAs with big tech companies to finance the uprate of existing nuclear plants to produce more electrical output.

However, a clear limitation to both these approaches is that there are only so many existing reactors currently in existence, and they may not be in the most perfectly suitable locations to build a co-located data centre or be viable candidates for uprates. Moreover, these approaches can be at risk of legal intervention.

The first approach described above, a PPA with an existing nuclear plant, was recently taken by Amazon at a site in the US, but the Federal Energy Regulatory Commission rejected an interconnection service agreement that would have provided power directly from the nuclear power plant to an Amazon data centre. Opponents to the plan had argued that the proposal would jeopardise grid reliability and increase rates for consumers.

Another option is to build a full-scale large nuclear plant from scratch in a location that would work for both a reactor and a data centre. The investment risk profile of this approach is higher, as full-scale reactors have often encountered cost overruns, delays and require large emergency planning zones.

Finding a suitable location, obtaining planning permissions and the strong likelihood of intervention by groups opposing nuclear all tend to make the timescale for new large nuclear plants on the order of decades rather than years. With developments in AI technology moving ahead at lightning speed, there may be hesitation toward an energy investment that cannot be rushed to match the same breakneck pace of tech development. In short, there is not enough time left to take this approach.

Which brings us to the third option: small modular reactors. These designs show a lot of promise, and so, unsurprisingly, have drawn a lot of attention from potential backers from the world of big tech. They promise to be built faster, come online quicker and be less demanding in terms of suitable location.

However, the small modular reactor (SMR) industry is one that is still developing. The UK is making good progress in reviewing a variety of SMR designs and the sector looks primed for take-off in the near future. But it is not in full flight just yet. That said, sizeable investment from the tech sector is a huge boon for any industry looking to get off the ground and could fuel a groundswell of confidence for other potential backers of SMR ventures moving forward.

Benefits beyond a cash injection

The potential flood of cash from big tech companies that this new interest in nuclear energy promises is stealing the headlines at the moment. However, there is perhaps a wider benefit that is being overlooked, but that could be just as profoundly impactful.

One of the key challenges that nuclear power – both full scale plants and SMRs – have encountered in the past decade is the issue of funding. The UK government has tried a few different approaches on this front, but it has not quite been able to find the right formula.

The contracts for difference scheme, which was applied to nuclear projects until 2021, required developers to front the capital expenditure for building reactors in return for fixed-price returns on the power generated once it was online. This model proved to be problematic and expensive, and so it was replaced by the regulated asset base (RAB) funding structure.

The RAB model sees consumers contribute to the construction costs, with the costs then returned in the form of energy savings over the course of the plant’s lifetime. This was introduced with the intention of reducing reliance on overseas developer investment, but would not be appropriate for data centre applications, because it is big tech that benefits directly from the power generated, not the consumer. Having the UK ratepayer subsidise foreign big tech companies would be a non-starter, so the RAB is off the table as an option.

The influx of private investment from the tech sector therefore brings the potential to add not just capital, but much needed innovation into how nuclear projects are financed. For example, by approaching small modular reactors with a ‘fleet’ mentality, where the objective is to produce as many projects as possible of an identical design, developers will be able to spread the investment risk of new reactors across the fleet, and the risk is not purely placed at the feet of the first customer.

In many ways, this proposed approach mirrors how commercial aircraft development and production is financed, as the first customer for a new aircraft type does not bear the full true costs of producing the first in a new series of airplane. Rather, the development costs are spread over the entire production run. If manufacturers are successful in making mass-produced small modular reactors in factories in large numbers, perhaps taking lessons learnt from the aviation industry could unlock delivery on a similarly large and reliable scale as airplanes.

Funding an entire fleet of nuclear projects is no small endeavour, but this is what will be needed to match the energy requirements of AI. The old ways of raising capital for these projects will not suffice, and so the truly exciting prospect of big tech moving into the nuclear investment landscape is that entirely new, innovative and much needed ways of thinking about how we finance nuclear energy projects could be introduced.

Once the deep pockets of big tech are on offer, the financial wizards in the City of London will likely apply the full weight of their intellect and creativity to financing nuclear.

 

Vincent Zabielski is a partner at Pillsbury Law. He can be contacted on +44 (0)20 7847 9537 or by email: vincent.zabielski@pillsburylaw.com.

© Financier Worldwide

BY

Vincent Zabielski

Pillsbury Law

©2001-2024 Financier Worldwide Ltd. All rights reserved. Any statements expressed on this website are understood to be general opinions and should not be relied upon as legal, financial or any other form of professional advice. Opinions expressed do not necessarily represent the views of the authors’ current or previous employers, or clients. The publisher, authors and authors' firms are not responsible for any loss third parties may suffer in connection with information or materials presented on this website, or use of any such information or materials by any third parties.