Data Center Thematics-Demand Driven From Artificial Intelligence AI
The data center thematic encompasses multiple industries including electric utility, electrical equipment and components, construction & engineering, semiconductors and last but not least energy.
Data center growth is projected at 11% going forward, largely driven by increasing artificial intelligence workloads. AI workloads currently account for 14% of global data center power usage, but this is expected to grow to 33% by 2030 atleast. The AI race among big tech and AI labs is as competitive as it has been this year. Every few months, atleast, there are new models being released that are better then the previous market leader.
Agentic AI is beginning to be commercialized by tech companies like OpenAI with their Operator. Google also released an agent throught their LLM Gemini. Anthropic introduced theirs also. So, the trend in AI growth is as clear as recent prior years. The AI demand is going to continue to need more data centers as time goes on.
There is also a turning point happening with mega data centers and the massive $500 billion Stargate joint data center project. The Texas data center site has buildings that are half a million square feet each. There are 10 buildings currently being built. They will expand to 20 other locations beyond the Abilene location. Construction is underway right now, and I believe this could be just the begininning of accelerating secular data center growth.
The Companies
Let's look at some of the major construction and engineering companies in this area and see what we can find. These large companies are getting atleast around a 1/3 of their revenue from data centers.
AECOM (ACM) is an engineering and construction company that currently has a record backlog and pipeline. They are beating earnings and raising guidance every quarter. They have growth in the highest margin markets. Free cash flow in the recent quarter was up 141%.
They've gotten data center contracts in the past internationally. In the U.S. last year they got the Databank data center in Culpeper, Virginia expected online in 2027.
Data center projects are higher margin than others for these kinds of companies. So, with my base case of accelerating data center growth, this is very advantageous.
While none of the large-cap stocks in this area are "cheap" with a trailing PE multiple of 23 for ACM, I see 12-15% EPS growth going forward. So, there is growth at a reasonable price (GARP) here with room for multiple expansion.
Jacobs Solutions (J) is a top data center construction and engineering firm. Just this month they got a contract with Nvidia to work with them on data center operations for the Omniverse Blueprint and Nvidia's AI factory digital twin blueprint.
They have a $22.16 billion backlog up a whopping 20% YoY. Jacobs has been buying back stock for years and share count has decreased year over year. I see them growing earnings 12-19% a year and the PE multiple is currently 22. Another GARP here.
Fluor (FLR) is another major steady player here, and they do hyperscale projects. The difference with them is they have a sizable oil and gas and chemicals segment. It's been a drag on growth for many years but is more stable the past five.
A different thing also is they have been a huge investor in Nuscale (SMR) that builds small modular nuclear reactors for power. At one time, in the past they had a $600 mil stake but have reduced this significantly just recently. They still have a sizable stake in the hundreds of millions and it's been dragging down their growth some. It's interesting because the market currently sees this as a negative as commercialization has been lacking but we all know the coming trend here with SMR's due to growing electricity demand.
Ten years ago energy used to be one of Fluors largest segments and it's been replaced with the urban solutions. With the recent White House order, just yesterday May 23, 2025 to advance national nuclear production and revitalize the U.S. nuclear industrial base, I see a likely scenario of the Nuscale stake vastly outpacing the oil and gas side.
Fluor TTM EPS is $2.57 and with the stock at $41 this is a 16 PE multiple. The lower PE seems to be a product of energy being about a 1/3rd of revenue and the margins here are very low and growth has been poor lately.
Bottom Line
I like all of these companies as long-term investments in the AI data center theme. I see market corrections as good opportunities to add. While other companies in the AI theme could offer more accelerative top line growth, I see these as safer stalwart type prospects.
There are many more data center investment opportunities I will post research on soon. So stay tuned. I recently did some analysis on data center REIT's Equinix (EQIX) and Digital Realty Trust (DLR). So, I think there could be the most long-term value in FLR valuation wise here vs others here assuming commercialization of SMR's picks up soon.