Tag: STOCK

Graphics processing units (GPUs) are among the most important types of hardware underpinning the artificial intelligence (AI) revolution are. Between them, Nvidia and Advanced Micro Devices essentially own the market for these advanced parallel-processing chipsets — data centers around the globe are outfitted with the wares of these two semiconductor specialists.

While that’s good news for Nvidia and AMD, there are other opportunities in the data center realm that I think many investors are overlooking. One such opportunity is Vertiv (VRT -3.83%), a stock that should really be on your radar as big tech’s investments in AI infrastructure continue to scale up.

What makes Vertiv unique?

One can imagine GPUs as being like a car’s engine; they provide the computing power that’s needed to train and run AI models. In that analogy, data centers can be thought of as the body of the vehicle. They are essentially enormous storage units that house vast arrays of server racks, each of which in turn is outfitted with loads of chip clusters.

The power those sites consume is enormous. According to a report from the Department of Energy, data centers accounted for approximately 4% of U.S. electricity in 2023. But it expects consumption levels to triple by 2028 — when it forecasts data centers will account for up to 12% of electricity demand domestically. One of the biggest factors influencing that rising demand? AI, of course.

But it’s not just processing that’s pulling all that electricity. A hard-working GPU server gets hot — and too much heat reduces chips’ performance and their lifespan. So servers and data centers have to be kept cool.

Today, temperatures are usually controlled in data centers through traditional methods such as fans and air conditioning systems. Vertiv provides an array of hardware for constructing data centers, but one area it specializes in is an emerging technology known as liquid cooling, and it’s gaining momentum.

The chart illustrates Vertiv’s revenue trends over the last several quarters. The slope of the company’s revenue growth is steepening at a considerable rate — but it’s where that growth stems from that has me most excited.

VRT Revenue (Quarterly) data by YCharts.

During the company’s third-quarter earnings call back in October, CEO Giordano Albertazzi said he was “very encouraged by the acceleration of liquid cooling revenue” and called it a “visible contributor” to the company’s recent growth.

Considering the company’s order book has increased by 37% over the last 12 months, I’m inclined to agree with Albertazzi.

Vertiv should benefit from AI infrastructure tailwinds

After companies experience phases of exponential growth, it becomes harder for them to impress investors. That said, I don’t think Vertiv has even hit its stride yet.

Over the last several weeks, a number of important announcements have been made related to AI infrastructure spending. For starters, OpenAI CEO Sam Altman joined Oracle‘s Larry Ellison and SoftBank’s Masayoshi Son at the White House shortly after President Trump’s inauguration to announce the formation of a $500 billion AI infrastructure project called Stargate. This news broke concurrently with Microsoft announcing an $80 billion data center project of its own, and Meta Platforms showcasing a $65 billion spending project on AI-related infrastructure.

I view the rise in capital expenditures from hyperscalers as a major tailwind for Vertiv in the long run. However, there is one big development to consider before scooping up shares of Vertiv right now.

Image source: Getty Images.

Is Vertiv stock a buy right now?

Over the last few days, you likely have been hearing about a new AI start-up out of China called DeepSeek. To summarize, DeepSeek built a generative AI model that’s meant to compete with OpenAI’s ChatGPT.

DeepSeek’s team says they trained their model using legacy chips from Nvidia rather than cutting-edge GPUs — a notion that has caused widespread chaos in the capital markets. In theory, if DeepSeek is as powerful as it claims to be, then new rival models could likewise be developed and powered using less expensive hardware. In which case, tech companies might not need to spend nearly as much capital as they expected to on the latest GPUs and new data centers.

How this situation will actually play out is as yet unclear. Among investment bankers, Wall Street research analysts, and technology enthusiasts, there are a host of varying opinions about DeepSeek and its capabilities. More so, there is a lot of conflicting reporting regarding how DeepSeek built its model. There is an existing scenario that the model was built using more sophisticated hardware than initially claimed.

The reason this is important is that the introduction of DeepSeek could inspire big tech to trim their capital expenditures (i.e., infrastructure budgets). Should this occur, I would expect Vertiv’s business to experience some form of deceleration as well.

Right now, Vertiv trades at a forward price-to-earnings (P/E) ratio of 30. That’s a bit higher than the average P/E of the S&P 500 (SNPINDEX: ^GSPC), which is about 24.

In a world where DeepSeek didn’t exist, I’d say that Vertiv was deserving of a premium, given the tailwinds from rising AI infrastructure spending. But given the contrary story lines and unfolding details regarding DeepSeek, it has gotten harder to predict how hyperscalers will spend in the near and medium terms.

For now, I think the prudent strategy is to listen to earnings calls from big tech and pay keen attention to their plans for AI infrastructure spending. From there, cross-referencing this information with the guidance Vertiv offers during its fourth-quarter earnings call in early February should help clarify what the company’s prospects look like.