The Perils and Profits of Riding the AI Energy Wave

Don't just yell cowabunga and get washed out

“AI is the New Electricity”

- Dr. Andrew Ng

“We see the wave coming…every company has to implement it — not even have a strategy. Implement it.”

- Emad Mostaque, Founder & Former CEO, Stability AI

Thanks for reading! This week, I’d like to invite all Dividends Premium readers to join me in the “Investing Channel” of our Dividends Roundtable. It’s easy to use and a great place to talk about stocks! I host it on our Discord platform.

To find out more about Dividends Premium and Dividends Roundtable check out the upgrade options highlighted in this email, as well as in the Substack app.

It’s no exaggeration to say Artificial intelligence (AI) is triggering a tectonic shift of continental proportions in our economy and society. Resulting tidal waves have made landfall in many areas. But one of the largest of them is still building.

That tsunami is AI's demand for energy. And the energy growth boom it's creating. The question of our time is: How can we ride this wave without getting drowned?—RC

The Massive Energy Demand of AI

AI systems consume staggering amounts of electricity. Processing AI queries requires 300% to 400% more electricity than standard internet searches. That shouldn't be a surprise. Search engines give users a pencil-thin search bar. Generative AI apps give you room to write dissertations for every question.

And all the tech giants are sprinting to further expand their AI capabilities. So as a result, their data centers are becoming energy behemoths. They could account for up to 9% of U.S. electricity consumption by 2030—more than double their current usage. This is exciting but fraught if your goal is to profit from all the commotion.

The Problems with Fossil Fuels, Renewables, and Conventional Nuclear

Why Fossil Fuels Can’t Cut It Alone

Fossil fuels have been dominant thus far. And they would be the clear choice if AI development were the only priority. But they are hamstrung. Environmental concerns have made fossil fuels a politically and socially unfavorable option. It's not just widespread public disapproval either. They also have higher costs compared to increasingly affordable renewable alternatives. And they're in conflict with Big Tech, who drive much of AI development. All the giants have already committed to ambitious carbon-neutral goals. 100% renewable energy usage and net-zero emissions puts a cap on conventional fuels.

Renewable Energy Limitations

But renewable energy sources such as wind and solar present their own set of challenges. The most devastating is that they are intermittent. Data centers need consistent and reliable power. They cannot afford the variability inherent in renewables. What's more, the grid is not currently equipped to handle the wild fluctuations of distributed renewable production. Batteries and other storage technologies need to be cheaper. Until then, renewables can't ride the AI wave alone either.

The Regulatory Hurdles of Conventional Nuclear Power

Nuclear would be the next obvious choice. It offers the most reliable and most carbon-free power.

Microsoft is well aware. They recently signed a 20-year power-purchase agreement with Constellation Energy. This was low-hanging fruit. It would completely avoid new construction and its prohibitive costs.

Constellation will now restart a reactor at Three Mile Island which it closed in 2019. Beginning in 2028, Microsoft would receive 100% of its 837MW of electricity. This would be the equivalent usage of 800,000 US households. Not so micro after all. Tellingly, the 20-year contract is significantly longer than Microsoft's solar and wind PPAs.

But the challenges of conventional nuclear plants are as large as they are. Regulators might not approve Microsoft's private power supply.

Amazon recently showed us the possibilities. And not in the sense of their cliche marketing fluff.

They wanted to secure a stable flow of Talen's nuclear energy to their data centers. But the Federal Energy Regulatory Commission (FERC) rejected their partnership with Talen Energy. The FERC said this would threaten the grid and raise costs for other consumers. No flow for the Amazon.

The Promise of Small Modular Reactors (SMRs)

Big Fixed Nuclear has problems. That's why Small Modular Reactors (SMRs) are emerging as a promising alternative. SMRs address many of the issues associated with traditional nuclear power. They provide scalability, reduced costs, and enhanced safety features.

BFN projects take a decade or more to construct, compared to SMRs which can take as little as 3-5 years. SMRs are prefabricated in factories. This reduces construction times and minimizes delays caused by weather or workforce shortages. And their smaller size is expected to makes them more affordable and accessible. Not to mention popular among developers and investors, who prefer less risky projects.

Data centers could theoretically add SMRs as needed to scale up operations. And for AI this is a huge advantage. Growth is difficult to predict but may come at a breakneck pace at any time.

The next concern with nuclear is safety. It has plagued its development for decades. If you've watched the excellent HBO mini series Chernobyl, it's easy to understand the fear. But when you compare the death count of nuclear to coal or even solar, it's hard to remain so prejudiced.

But these concerns could be further addressed by SMRs. Passive cooling technologies and automatic shut-off features reduce meltdown risk. And their underground installation reduces the risk of accidents, natural disasters, and attacks. They also take up less space and use less water than traditional nuclear.

It's no wonder tech giants are already embracing SMRs to meet their energy demands. Amazon is partnering with Dominion Energy (NYE: D) and X-energy. Google plans to purchase energy from SMRs developed by Kairos Power. And Microsoft is giving more attention to SMRs as well. And they should. Bill Gates has been working on SMR tech through his company TerraPower since 2008.

The challenge for SMRs: At this point, it’s all in the prototype stage. There are no designs that have proven themselves at the commercial scale. That means any discussion of what they may ultimately cost and the time it may take to build is still theoretical. And so long as that’s the case, we’re unlikely to see commercial orders, no matter how much money Big Tech is willing to spend.

How to Ride this Wave as an Individual Investor

The principle: look for the "picks and shovels" of this new gold rush. They are the utilities and energy providers.

Utility companies play a critical role as the backbone of the AI energy revolution. They own the energy infrastructure AI data centers need. And their regulatory frameworks position them to profit from AI's increasing energy demands. Also, unlike Big Tech, they are often undervalued. If you can invest long-term, they are an excellent vehicle for profit.

Likewise, as energy demand grows, stable providers will profit. This means conventional nuclear, natural gas, solar & wind providers, and eventually SMR operators.

The AI energy wave represents a chance to align portfolios with the future of energy.

Focus on utilities and the nuclear renaissance. These are the picks and shovels. No one can predict who will strike the AI gold mines.

But they all need more and more power, and they need it delivered at scale across huge distances.

One important caveat: Ignore anything you read that attempts to convince you that there’s a perfect energy source. Chances are there’s a pretty hefty dose of self interest at work in those arguments. And frankly, anyone with any industry experience can tell you there’s no such thing. All of them have their challenges and limitations as well as strengths.

The best in class US utilities have known for years that the best “power mix” is a diversified one. And they’ve put that into practice delivering electricity that we’ve come to take for granted, even as it’s become more critical to our lives.

The challenge of providing enough energy for the AI revolution is no different. And the utilities and power producers that will give us the best returns—and build real wealth from the AI revolution—are going to be those that have a diversified power mix that combines the cheapest and easiest to build (solar, natural gas) with the largest and most reliable 24-7 (nuclear). Even coal will continue to play a role, bridging the demand gap between supply and rapidly rising demand.

One great example: Brookfield Renewable Corp (NYSE: BEPC). It’s a global leader in hydropower, wind, solar and storage. It owns 50% of Westinghouse, the leading US developer of both large nuclear (AP1000) and SMRs (AP300). And it pays a generous yield to boot.

Surf's on!