Full of energy: Capital Power’s Avik Dey is at the centre of an electrical storm in Alberta

Banks aren’t the only businesses in this country that stretch back more than a century. For nearly as long as electricity has been distributed through wires, Capital Power Corp. has profited from it. Of course, back in 1891, it was a utility called Edmonton Electric Lighting and Power. But then, as now, the flows of current and currency were entwined. So it’s fitting that since 2023, Capital Power has been run by a money guy.

Avik Dey might dispute that characterization. But throughout his career, from his CFO and investment days in Houston’s oil and gas market to the seven years he more recently spent running energy and resources at CPP Investments, Dey has been tied to the finances of energy.

Now he’s president and CEO of Canada’s largest independent power producer, generating 12 gigawatts via a 35-plant fleet. That puts him at the centre of an electrical storm in Alberta, where the push for data centres to capitalize on its abundance of power is meeting resistance from worried rate payers. To understand the dynamics of power production right now, we reached Dey in Calgary.

Can Capital Power lure AI hyperscalers to build big in Alberta?

You came to Capital Power in 2023. What was your mandate?

When I joined CPPIB in 2014, we had this thesis that (a) traditional energy still mattered; (b) the world was going to decarbonize; and (c) energy demand was going to continue to grow. Within that context, we felt it was important to be invested in critical assets. When I came to Capital Power, it was an extension of that same thesis. I saw it as a diamond in the rough, an organization that was incredibly good at running utility-scale assets. And I had a view that we needed more of it, not less—you could generate strong returns off a platform like this by growing and scaling it. That’s what I thought coming in.

Did that change?

Well, the underlying thesis was, the grids across North America were becoming more and more unstable. They were more and more unable to meet peak and baseload electricity demand because we were retiring old, inefficient, high-emitting coal plants and natural gas plants that provided baseload dispatchable power, and introducing more and more renewables. So the view was that we couldn’t turn those assets off, because demand for electricity is going up from things like GDP growth, automation, electrification. Couldn’t even spell AI at the time. What AI has done is it’s dramatically accelerated that and also made it a kitchen-table conversation, where everyone’s now concerned because it’s such a high and large load.

Capital Power’s assets are 60% in the U.S. and 40% in Canada. What should we take from that balance?

We don’t have any allocation objective by geography, very explicitly. Our expertise is natural gas. We go into those markets that require natural gas for baseload power. If a jurisdiction is well encumbered with hydro or has a long history of nuclear, they may not necessarily need natural gas for baseload. When we think about our diversification, we’ve got five key markets around North America: Alberta and Ontario in Canada; PJM, which is eastern U.S.; MISO, which is the Midwest; and WECC, the western U.S. All those markets require natural gas capacity. All those markets have access to natural gas feedstock through pipelines or reserves.

How has the investment landscape changed in Alberta recently?

It’s been on a path to change over the past three years. Speaking specifically to electricity, Alberta is one of two markets in North America that are unregulated. We call them energy-only markets, which means it’s private investment that facilitates the investment in new generation. As we introduced high levels of renewables, and it became more and more unstable, we needed to tweak that system to encourage new investment and balance the integrity of the overall electricity system, and in addition, leverage the infrastructure that’s built here to accommodate the AI.

Where do you see renewables fitting in for Capital Power and Alberta in the near to mid-term?

We continue to believe we’re on a path toward decarbonization over time. But you have to do it affordably. You have to meet reliability standards. And you have to have an eye toward what’s ultimately sustainable. So we articulate it as basically a 70-20-10 capital allocation strategy for our company—70% being gas-fired generation, 20% being renewable.

What’s the 10?

The 10 is technology and innovation, and supportive trading. Those are things that allow you to balance everything. I can provide you natural gas–fired power generation that meets a minimum reliability standard. But I can augment that with solar or wind. I can also provide reliability through battery storage. I can support that by creating energy risk management and trading environmental credits, or trading the underlying natural gas exposure, or providing energy risk management on power exposure over a short, medium or long term. Even though I talk about natural gas as our core competency, we have an active solar development program. We have utility-scale batteries installed. But you have to look at it from a system perspective. You need all of the above.

Is power production growth in Canada all about AI now?

The sources of electricity demand growth are multifactor. It’s GDP growth, it’s automation, it’s electrification. The difference with AI is, bringing a very large data centre requires hundreds of megawatts of power. AI is a forcing mechanism for the conversation on how do we build reliable grids, how do we make them affordable, and how do we address some of the system constraints that currently exist? Almost 75% of the grid in North America, or specifically the U.S., is aged out, or needs replacing or upgrading. If I have one project that’s 500 megawatts or 1,000 megawatts, it’s easy for us to talk about it, because you know who the counterparty is, and we can address what the system costs or benefits are by having them connect into the grid.

You’ve quoted Samantha Dart from Goldman Sachs saying we’re not adding capacity fast enough. What’s getting in the way?

In Canada, I think the ambition is there. It’s how do we convert ambition to execution? We’ve got differing systems in each province. We’ve got different market structures. What’s missing is the ability to build large infrastructure, whether it’s permitting, regulatory, attracting capital or providing certainty for investors. I think the last few months have been great in terms of announcing some large projects. But how do you turn it into an executable project?

There are other bottlenecks. Turbines are an issue, for instance. Can you explain that to me?

Turbines are basically the engines that power these power plants. In most electricity markets, if you need to build new generation capacity, you apply to the rate commission to have your project approved, it’s approved at a regulated rate of return, and then you go order turbines. In the late 1990s and early 2000s, we built up a bunch of natural gas–fired generation capacity. Through the 2000s and 2010s, we were retiring a lot of that capacity with renewables. As a result, there’s a very limited supply chain building and assembling that equipment. Over the past five years, a number of large players have reserved all of those forward slots to take deliveries of turbines. If you want to build a new large power plant now, you literally cannot access turbines unless you put an order in for 2032, 2033 delivery.

Canada still needs to remove regulatory roadblocks, Capital Power CEO says

Let’s talk about the Meta deal. How did that unfold?

We didn’t bring the project to bear, so I can’t speak to the specifics, but ostensibly it’s been in the works a couple of years.

But you’re providing it with power.

Let me just step back. So Meta agreed with Pembina and Kineticor to build new generation in support of their data centre in Sturgeon County. Our role is, we’ve provided a 250-megawatt-plus, 10-plus-year contract to sell them virtual power. The way the system works in Alberta is, anyone and everyone connects. And when you pull power from that grid, then you can price that power. It doesn’t have to be a physical flow of power to sell it. We’ve provided Meta with power price certainty over 10-plus years.

The province talks about “Bring Your Own Power” proposals for data centres. What are the dynamics of those?

When the AI super-thematic blew up, there were two types of power solutions. The first one was called BTM—behind the meter. Think of that as ring-fenced from a connection to the grid. That was ideal for everyone, because you didn’t burden the ratepayer. The challenge was, you couldn’t provide what’s called “five nines reliability,” which is 99.999% reliability. If I had a 1,000-megawatt plant supporting a 1,000-megawatt capacity data centre, I would have needed 1,400 megawatts to supply it because of the ancillary power. Then I would have had to double that generation to nearly 3,000 megawatts to ensure I had enough capacity and reliability to meet the five nines.

Solution two was, allow me to connect to the grid so that it provides that overall reliability backstop. Then I can more closely match, one to one, the generation requirement. The No. 1 criterion for hyperscalers is speed. How quickly can I get connected so that I can start processing that AI compute? Well, I can build a data centre in 24 to 36 months. The power solution could take five or six years. So where can I build the data centre, connect and get the necessary reliability? That’s where the market has gone. That connection to the grid provides the fastest and most reliable path to power.

But that affects prices for regular customers.

Exactly. If you get that grid connection but don’t bear the cost of it disproportionately to the average consumer, then you’re driving up the cost to the consumer. The conversation that’s occurring across North America now is, we have this really large customer that needs access to the grid, but there’s a potential net benefit here, because it can help pay for the upgrade of the overall system.

There’s a growing resistance to data centres. A Léger survey in July found that 81% of Albertans are worried data centres will spike their electricity bills. How concerned are you?

I think we have to be very concerned. It’s incumbent upon us to engage and partner with stakeholders. I don’t think these projects can go forward, given the size and scale of them, unless there is support from the communities. I think with AI specifically, it’s a large unknown. Ultimately, you have to be able to demonstrate that these projects are of net benefit.

I want to talk about your Genesee site. You see it as a crown jewel asset. What makes it so?

It was the site of a coal mine and a coal-fired power plant. It is today the most efficient natural gas–fired power plant in all of Canada. We have over 23,000 acres. We have strong relationships with the community. We’ve got high-voltage transmission at the location. We have access to water. And the site is just over 50 kilometres southwest of Edmonton, close to the airport, close to fibre infrastructure. Because the province and ratepayers have already paid for the infrastructure to connect that site, we’ve got significant and material expansion capabilities there that ultimately will benefit the ratepayer because we’re using largely existing infrastructure.

What are you hoping to do with it?

We’re hoping to build new generation that will contribute to the grid, for anyone and everyone who decides to build data centres or a new fertilizer plant or a new refinery or a new petrochemical facility. Or we could co-locate a data centre on our land and provide energy directly to it, and additional generation into the grid.

You recently partnered with Apollo Global Management. It put up the majority of $3 billion for the acquisition of some facilities. What does that deal do for you?

We take pride in the fact that we are investment grade. Our business takes underlying commodity-price exposure risk, because our input cost is the hydrocarbon, and our output cost is the electricity price. One of the foundational tenets of our strategy is managing how much we contract at fixed prices versus how much merchant risk we take on commodity and power prices. You asked me about geography; the more important mix for us is managing contractedness versus merchant exposure, which is 60-40. We continue to see a growing opportunity to acquire merchant assets that are predominantly not contracted, because we think we could buy them, improve them and contract them over time. But we didn’t want to compromise our investment-grade rating or materially shift our business risk. So the opportunity with Apollo was to pre-wire a partner that had confidence in us to source, evaluate and manage these assets, and then improve our competitiveness to go acquire more. Predominantly U.S. merchant natural gas plants.

In March, you said, “It’s the decisions we make in the next few months that will determine whether Canada seizes the moment.” Did Canada seize the moment?

We’re in early innings. All indications are yes. And we’re catalyzing around the necessity to build infrastructure. The ambition is not enough. We’ve got to push into execution. I’m excited about where we are. It’s the first time in a generation where I feel like, from sea to sea to sea, Canadians are catalyzing around the necessity to get on with it and do it.