Dell COO Jeff Clarke on Soaring AI Demand and Why It’s Still Early in the AI Revolution

Oct 08, 2025 15:50:00 -0400 by Tae Kim | #AI #Barron's Tech

Jeff Clarke, Dell’s chief operating officer, spoke to Barron’s about soaring demand for AI, use cases for the technology, and the challenges going forward. (Courtesy Dell)

This article is from the weekly Barron’s Tech email newsletter. Sign up here to get it delivered directly to your inbox.

AI Use Cases. Hi everyone. Dell Technologies has become a beneficiary of the artificial intelligence megatrend by giving corporations the hardware they need to take advantage of the technology.

Dell has built its AI server business to a projected $20 billion in revenue this fiscal year, from nearly zero two years ago. The company thinks we’re still in the “early innings” of the AI revolution.

This week, Dell’s stock hit a new 52-week high after the company boosted long-term financial guidance at its financial analyst day, citing strong AI demand. Dell is now targeting annual revenue growth of 7% to 9%, up from a previous range of 3% to 4%. Its adjusted per-share earnings growth target was boosted to 15% or better, almost double its prior forecast of 8%.

Dell is benefiting from a global distribution and services footprint. CoreWeave and Elon Musk’s xAI are both customers of Dell servers for their AI infrastructure buildouts.

Barron’s Tech spoke with Jeff Clarke, Dell’s chief operating officer, at the end of the company’s analyst day event in New York on Tuesday. We talked about how AI is being used, the state of AI demand, and the potential challenges going forward.

Here are edited highlights from our conversation:

Barron’s: Let’s talk about AI demand. Where are we in the AI demand curve, where are we going and what is driving it?

Jeff Clarke: It’s ultimately materialized at a much faster and greater rate than we thought. Two years ago, for 2025 we thought AI capex spending was going to be $200 billion. Now it’s more than $400 billion. For estimates on AI hardware and services by 2027, it was going to be $124 billion, and now we think it’s greater than $310 billion. Those are reference points of the acceleration of AI.

For drivers, we had gen AI, one-shot models, reasoning, moving to agentic autonomous agents, and physical AI. The message is we are still early on in AI. The early innings, if you prefer that analogy. We’re early in the discovery of the applicability and how capable this technology is and what it can do for customers.

AI is helping them with developing software code. It’s helping them with fraud detection or anomaly detection in IT systems. It is accelerating drug discovery. These are real practical uses.

You have talked about how Dell is using the company’s own positive experiences with AI to close new sales deals. What is Dell doing with AI?

What we’re seeing is our developers are more productive. They’re developing more code faster and at higher quality. That’s a pretty darn good outcome.

And in the area of services, we have found many opportunities to bring lots of disparate data that we had: call data, case data, dispatch data, repair data, etc. We actually used AI to bring that data together to simultaneously mine all the data sets. We are able to translate that into a service assistant that sits on the shoulder of every one of our tech support folks. We close calls quicker, we dispatch fewer parts, and we’ve increased customer satisfaction.

The one that I’m probably most excited about now is the rolling out of what we call a Dell sales assistant. We’re using AI to take all of our internal data and all of the appropriate external data, and put it at the fingertips of every salesperson.

It makes what you do better. Better decisions, faster decisions, greater insight, better pattern recognition. And who doesn’t want that? Whether you’re selling something, building something, designing something, or servicing something.

What about the competition? Why are you winning these deals with CoreWeave and xAI?

We are differentiating ourselves on superior engineering and engineering execution, our ability to take an idea, create a design, and to deliver it in a very short period of time.

We were the first to market with Nvidia’s GB200, the first to market with the GB300. When our gear is delivered, it’s up and operational in 24 to 36 hours, which is differentiated in the marketplace. We did a 110,000 GPU deployment, liquid cooled, from design to delivery in roughly six weeks.

Then lastly, we would say one of our inherent advantages is we are able to provide flexible financing to customers. That’s why we think we’re winning. We’re not the low-cost provider.

Talk about the next few years. With Nvidia’s GPU road map and its higher power density, what challenges will the industry face?

The density of power and the density of GPUs that are going into a rack, and ultimately into a data center, continues. Not too long ago, a standard data center would have 10 to 12 kilowatts of power on a square where the rack sits. Our first big rack scale Nvidia Hopper designs were 100 kilowatts. Now we’re regularly over 200, heading toward 500 and then toward a megawatt.

The GPU density is scaled with that, and it just drives incredible power cooling and software management challenges.

It’s an opportunity for us to differentiate. We’re doing some of our own cooling design. We’re working with partners to figure out new materials for heat transfer from the GPU and get it out of the rack itself, and dealing with the energy density. On power, how do we get more power efficiency? We’re looking at new materials, like gallium nitride and silicon carbide, to drive new energy density and energy efficiency.

Tell me about the sovereign AI business, selling to governments. Where is it now?

Sovereign is a big opportunity. It continues to mature. The pipeline continues to build with sovereign opportunities. I think they have very similar characteristics to some of the large tier 2 cloud service provider customers. We’re excited. With the U.S. government, we look at that as a sovereign opportunity. We had a win with the Department of Energy. I think there’s more to come.

Thanks for your time, Jeff.

This Week in Barron’s Tech

Write to Tae Kim at tae.kim@barrons.com or follow him on X at @firstadopter.