Among recent AtkinsRéalis actions to expand its nuclear market share are seeking a U.S. Nuclear Regulatory Commission license for its enhanced CANDU pressurized heavy-water reactor design, and collaborating with AI computing giant Nvidia to design and deploy large-scale, nuclear-powered AI data centers. ENR Deputy Editor Debra K. Rubin late last month interviewed AtkinsRéalis President and CEO Ian Edwards, in the position since 2019, and U.S.-Latin America President Byron Bright, who joined in June from a 15-year career at KBR, most recently as chief operating officer, for insights on nuclear market strategy and corporate changes over a decade. This interview has been edited for clarity.

Ian Edwards: Obviously, there’s a massive renaissance in nuclear power. We’re seeing it globally, with 38 countries that have now signed the COP 28 pledge [in 2023 to triple nuclear energy capacity by 2050]. That is real. I visit energy ministers all over the world, and you now see it in numerous energy policies. We’ve actually been rebuilding the CANDU nuclear fleet for the last decade. We’ve rebuilt about 10 reactors in the world. We give them another 35 years of life—and are doing it on time and on budget, which are not words normally associated with nuclear construction. We’ve also gotten our first new build orders for two reactors in Romania, and are working closely in Canada to get more.

Edwards: What is transpiring more recently, this year particularly, is this hyperscale opportunity to put nuclear power into a data center for data storage. Because obviously the key constraint is rapidly becoming electrical energy, hyperscalers [are] looking to develop gas connected to data centers, but we can now offer comparable cost and even time in nuclear power. Our reactor is a 728-MW pressurized heavy-water reactor that uses natural uranium fuel [and] is quite easy to build. Combining an engineering services business with a nuclear business gives us that scale to deploy for nuclear customers. The plan would absolutely be engagement with utilities, which we have been doing, and with hyperscalers for potential behind the meter developments. There’s no specific opportunity to talk about today but we’re working with numerous organizations to secure new build opportunities. I would say if we identify one or two sites or one or two customers, there’s a lot of work to do before execution with needed project specific approvals, environmental impacts and state licensing. So clearly with that kind of opportunity we are exploring nuclear power data centers with Nvidia [to provide a framework to design and optimize integrated infrastructure systems in a 3D digital twin] before building in the real world.

Edwards: We want to be ready with the licensing. A few weeks ago, after a period of consultation with NRC, we decided to make the formal application because we believe the technology is licensable to deploy in America, with power demand so high. It’s not about competing with the Westinghouse AP-1000 design, the American technology. The issue is does the industry have the capacity to deliver what’s necessary for electrical energy? Capacity is going to be the differentiator of companies. Eighteen months is the timeframe under Article 53, [the new NRC rule to accelerate and simplify the advanced reactor licensing process while continuing to ensure safety]. CANDU is a fairly simple technology that is heavily regulated in Canada, so we believe we may be able to beat 18 months.

Byron Bright

AtkinsRéalis' very substantial engineering services business covers all major disciplines of the project lifecycle from early planning through design through operation. In the U.S., we’re building that out pretty aggressively. We’re very strong in the Southeast and we’ve made acquisitions in the West, so we're really growing across a broad portfolio of infrastructure to include power and energy and transmission and distribution. We do need to do a lot more work on our name in the U.S. because the brands are not as recognizable, except in certain states like Texas and Florida. We have an opportunity now in the U.S. to really highlight how differentiated we are and become a top player. There’s a desire to be successful and deliver across the portfolio, but also a humility that we don’t know all the answers and can work with partners and local supply chains.

Edwards: Most of it, but not on the nuclear side where we have grown organically. Our nuclear business three years ago was 10 to 15% of the whole company. Today it is 25% and its going to be 40%, because both it and engineering are growing. Nuclear will grow faster, it grew at 50% in 2025. But building our engineering capability will come from acquisitions. We’ve done two in the U.S. and have a pipeline of opportunities to acquire more. At the moment, we’re ranked at No. 16 on the ENR Top 500 Design Firms list, but that is not where we belong. A company of our scale needs to be in single digits, and we have a plan for that. We’re quite successful in the U.S. in transport and general infrastructure, but our place to grow here is in nuclear.

Edwards: Our businesses in the U.S. will have U.S. people. If we build a nuclear plant here, our intention would not be to bring over a whole bunch of companies from Canada. It would be to develop a supply chain in the U.S. There may be some manufactured components that would potentially suffer from tariffs, but the intent would be to build a U.S. business. We’re very local—American in the U.S., British in the UK, Canadian in Canada. We have people in just about every state who understand local permitting laws working with local clients. There’s a lot of public relations that go wrong around data centers and nuclear. We support a lot of customers today doing early works, prefeed and concept studies. We want to build U.S. jobs here and continue to grow that workforce. Demand in the U.S. won't be met unless some other kind of technology comes into the market. As long as we develop the U.S. supply chain, make sure we have a U.S. business and minimize the amount of goods shipped across the border, it’s going to be OK.

Edwards: Reactors in Canada are largely rebuilt, but we’ve had to reskill a lot of people in the ecosystem—universities, supply chain and workers, including our own. We had 3,000 nuclear folks two and a half years ago. Now we’ve got 6,500. A lot are new graduates, but most of those are people taken from engineering in another field and given extra skills. One advantage is that the planning period for a nuclear plant is minimum two years, probably three. So that’s the time to build up your skills and your retraining. We’re working on major projects around the U.S. with lots of general contractors on the construction side, or other partners. The U.S. has a broad ecosystem we will pull from as the projects come. We have the experience of what the long lead items and bottlenecks are.

Edwards: We’ve done eight reactor upgrades and put them back on line in the last decade and there were a couple done before that. The first one was always the most difficult because engineers and craft labor are doing it for the first time. By the third, we were working ahead of time and on budget. Because the industry has been fairly flat after the [2011] Fukushima earthquake, until recently, there’s not a lot of companies that have had the experience we’ve had with these rebuilds. We also have quite a large transmission business mainly out of Canada but also do that work in the U.S., UK, Middle East and Australia. We used to own a transmission network in Alberta [Alta Link], but sold it in 2014; we sold the asset, but kept engineering credibility, and demand is very high. I wish we had more people and capacity. This could be a major growth area.

Bright: I did a lot of flight test engineering [as a U.S. Air Force flight test engineer], which taught me how to work with exceptional teams and in high consequence environments. This translates well to AtkinsRéalis, in mission critical work. Skills I learned early in my career have led me to where I can be part of something bigger and help make the right decisions.

But it doesn’t matter if you’re Westinghouse or CANDU, there wont be enough capacity to meet demand. It’s a big market and there’s room for all players and multiple technologies. If the U.S. really wants to be premier in AI and data centers, it's going to need all the sources of power it can get built in a reasonable time frame—and nuclear is a huge part of that. I think some tech hyperscalers possibly will wait a bit to get the nuclear. My impression from them is that they’re talking about [needing] 10 times the power they have today, so that’s a challenge. Technology will improve such that power demand comes down, but there's still an awful lot of scaling up to do. Hyperscalers realize they have to go behind the meter—in other words, do their own power.