Data centers are experiencing an unprecedented growth in demand. To understand how data center owners and operators across the United States are planning capital investment around power demand, we caught up with Andrejka Bernatova, Founder, Chairman & CEO, Dynamix Corporation for this exclusive tête-à-tête.
Q-1) In your opinion, what is the single most defining feature of investments in energy and power for data centers in the United States?
The defining feature is the gap between projected power demand, and the infrastructure that can realistically deliver it. The bottleneck has shifted from capital to power. The market is filled with “PowerPoint gigawatts.” Announcing new generation is easy, but delivering reliable power is much harder.
That gap is what separates one project from the next. Many of the data center projects being announced today will never be built unless developers secure power, finance the infrastructure and get customers to make long-term commitments. Until then, those promised gigawatts may never materialize or, at the very least, may take longer to deliver.
Q-2) How are investors assessing the risk-return profile of energy infrastructure investments supporting AI and data center growth in the United States?
The real test is whether developers have the contracts, financing and ability to deliver what they promised. I look past the headline demand and underwrite the project itself. I want to know how far along the site is, where it stands in the permitting and interconnection process, who will build and operate the generation, how fuel will be secured, and whether the customer has made a long-term commitment.
A large contract with a hyperscaler does not automatically make it a high-quality business. You still have to get the infrastructure in the ground and operating. Some of these large-scale power projects supporting AI demand will become long-term winners and others will end up being zeros. Investors need to be selective and also more involved, roll up their sleeves and help execute projects of this scale and complexity.
The scale is also why I believe AI data centers will become their own investment class over the next five to ten years, on par with midstream and upstream energy. The demand is so enormous that this infrastructure should be built and financed separately from the hyperscalers.
Q-3) Are captive power generation projects for data centers drawing more attention than investments in existing energy companies that also supply power to data centers? Why or why not?
I see opportunities in both. AI is changing the order of operations for data center development. Power has to be designed into the project from the start, not addressed after the site and facility are already in motion, and dedicated generation is how many developers are trying to do that.
Existing energy companies have an important role here. Data centers need power, and that power does not come from thin air. Gas-fired generation depends on midstream infrastructure for its fuel supply. That is why I expect many traditional infrastructure companies to move into the power space. Traditional infrastructure and generation together form the backbone of how data centers will secure power.
Captive generation also has an upside beyond the data center itself. Where its grid connection and market rules allow, a data center with its own generation could sell power back to the grid during peak demand or extreme conditions, like winter in Minnesota or summer in Texas, helping stabilize supply when it is needed most.
Q-4) Are traditional infrastructure investors entering the energy sector, and are they pricing these assets as long-term infrastructure investments with return expectations comparable to regulated grid infrastructure, or as higher-risk investments requiring a higher return premium?
Yes, and I think that is a good thing. Scaling this kind of power infrastructure takes investors with deep capital reserves and real experience deploying capital in capital-intensive and highly complex projects. High-quality players, especially the ones backed by infrastructure funds and private equity, are continuing to deploy. Smaller players can also play a key role, using their nimbleness and speed to create a strategic advantage.
On pricing, I do not think these assets belong in one bucket. Contracts in connection with the ability to execute are the most important piece. One without the other is not effective in isolation. When the customer is willing to sign a long-term contract and that is supported by execution ability, it supports a strong financing environment. But the terms matter. A poorly priced power contract can leave you facing bankruptcy or steep price increases a year or two later. Having a contract does not eliminate execution risk.
This is the same lesson we saw in clean energy.
Q-5) How are energy and power investors factoring in data center protests and their possible adverse impact on projects into their investment strategies?
Investors are watching local and state pushback very closely. Six to twelve months of prolonged opposition can push developers to more favorable jurisdictions.
The concerns on the ground are real. I have visited these communities and understand why people are pushing back. Promises of jobs and investment need to address residents’ concerns about electricity costs and water supplies. I saw this play out in oil and gas, where the industry had to learn how to bring communities along as partners and data centers are showing some of the same elements.
Investors are mainly looking at whether the power developers can actually deliver. We are seeing a lot of companies that do not have the experience to navigate permitting and supply chain constraints, so the question is what the path to execution looks like. Blanket moratoriums will not solve the underlying grid problems, but communities are right to ask who will pay for new generation, grid upgrades and the pressure these projects place on local power and water supplies. Hyperscalers cannot expect everyone else to pay for their power demand.
Q-6) In the US, which emerging renewable and low carbon technologies are currently most attractive to investors?
A diverse energy mix is critical to resilience. The Dakota Access Pipeline, Mountain Valley Pipeline, and LNG permitting bans under previous administrations show that even projects with tens of billions of dollars invested can be delayed or shut down when political priorities shift. Projects without a clean energy component may face greater risk. Natural gas is a short-term fix, but it must be paired with other solutions on a robust basis.
I would start with what can actually be deployed earlier, because the real pressure is on the next five years, not some distant horizon. Solar and wind paired with batteries are easier to implement than most of the alternatives, and renewables will keep expanding regardless of which administration is in office, simply because that is where the market is moving. But wind and solar cannot meet these power needs on their own. They need support from reliable generation, storage and microgrids.
Geothermal and carbon capture are getting a lot of the excitement. Nuclear could play an important role, but new nuclear projects do not solve the immediate demand problem, and execution, permitting and community resistance are real hurdles. I think valuations in nuclear are inflated relative to what can realistically be delivered in the near and medium term, and the enthusiasm reminds me of the hydrogen euphoria. Great concept, but impractical at the projected scale and speed.
Q-7) What opportunities and trends do you foresee in 2027 in investments into energy and power for data centers in both Europe and the United States?
In the US, the decisions being made right now on the energy mix will shape 2027 and well beyond. Once billions have gone into gas pipelines and generation, it becomes much harder to come back in five or ten years and fit other sources around what has already been built. Gas is an essential part of the answer, but companies like Microsoft and Google still need a more balanced energy mix as a resiliency tool for the project and potential regulatory scrutiny if they only have a fossil fuel solution.
The conversation has gone from climate commitments to “give me natural gas because it is the fastest and most resilient option.” I know hyperscalers need gas, but, for example, if they are considering 2 percent geothermal, they should ask what it would take to get to percent. We need fossil fuels, but we need real steps toward a balanced energy mix especially given the volatility of regulatory environments globally and in order to address community feedback.
Texas is the other thing I would watch. The question is which projects can secure power and move forward, and whether local pushback sends developers elsewhere. Price volatility matters too. When energy is volatile, it is tough to build these gigawatt platforms, and with policy that can reverse every four to eight years, investors will want a resilient energy mix.
In Europe, the story is energy security. The need to diversify gas supplies opens up opportunities for LNG exporters, along with the pipelines and facilities needed to make that happen. My view is that it is not realistic for Europe to materially reduce its reliance on US energy without creating problems for security or affordability. Europe needs a healthy relationship with US suppliers. Nuclear will stay part of the conversation, but it is very hard to see new nuclear projects delivering at scale in the next five years.