AI heavyweights like Meta (META), Microsoft (MFST), Alphabet (GOOG), and more are spending billions of dollars to build data centers and other infrastructure necessary to support their AI ambitions.

As the big players race to outbuild each other, capital expenditure on data centers are expected to surge to nearly $7 trillion by 2030.

However, data center buildout is only part of this complex challenge to achieve AI’s potential.

While hyperscalers are racing to develop more advanced models, a second race is taking place to find ways to power data centers and servers that support AI. 

Back in 2022, the Federal Energy Regulatory Commission (“FERC”) forecast that peak energy demand in 2029 would rise 23 gigawatts (“GW”) from 2022 levels to 840 GW. A year later, FERC revised its estimate, projecting peak demand of 859 GW by 2029.

And since the proliferation of AI and the data centers which power them, FERC has once again increased its energy demand estimation to 947 GW by 2029. 

Furthermore, data centers are forecasted to consume as much as 12% of electricity demand in the U.S. by 2030.

With AI becoming an increasingly significant drain on the U.S. energy grid, many industry experts are pointing to nuclear as the long term solution to growing power demand.

Bloomberg estimates that there will be $350 billion invested into the nuclear industry through 2050 as new reactors and small modular reactors (“SMR”) are built over the next decade.

The resurgence of the nuclear industry will unlock opportunities for related industries, namely for companies specializing in nuclear safety and monitoring.

Mirion Technologies (MIR) provides nuclear measurement and detection systems for nuclear reactors.

The company’s offerings include radiation detection systems such as dosimeters, reactor monitoring solutions, and decontamination support systems, all of which are critical to a nuclear reactor’s operating life cycle, from construction to operation to eventual decommissioning.

Mirion’s technology is used in over 95% of reactors worldwide, giving it near-monopolistic control over the nuclear monitoring industry.

With demand for nuclear energy on the rise, Mirion’s business is showing signs of growth. Between 2018 and 2020, the company’s revenue grew around 3% per year, while this has picked up to 9% per annum since 2021.

What’s more, Mirion expects revenue to grow between 6% and 8% annually going forward. 

However, while other AI-related stocks have seen valuations skyrocket recently, investors may be underestimating Mirion’s potential due to as-reported accounting distorting the company’s true profitability.

Mirion hardly looks like an industry leader serving a vital role in the nuclear space based on GAAP accounting.

As-reported return on assets (“ROA”) has averaged less than 1% since 2019. It peaked at just 1.6% in 2021, and returns were back below 1% last year.

Meanwhile, Uniform Accounting portrays Mirion has a profitable business, reflective of its leadership status.

Its Uniform ROA has been above 20% in five of the past six years, generating 24% returns in 2024.

As nuclear power continues to see a resurgence in demand, Mirion Technologies stands to benefit as its offerings are critical to the industry as a whole. And with investors misunderstanding how profitable this business is, Mirion could present an opportunity for investors looking to capitalize on markets running parallel to the AI boom.

Best regards,

Joel Litman & Rob Spivey
Chief Investment Officer &
Director of Research
at Valens Research

Today’s analysis highlights the same insights we share with our FA Alpha Members. If you want to an get in-depth analysis of market trends and uncover undervalued stocks, become an FA Alpha Member today.