Dell COO: AI will move to the data
"We believe the long-term AI action is on-prem"
As generative AI training and deployments move from text to richer media and other unstructured data, there’s a huge opportunity for storage and professional services growth, says Dell Chief Operating Officer Jeff Clarke.
Much of that will be in on-premises or at the edge, he believes.
And in his view, Dell is “looking at an opportunity where every dollar that is for a AI server, GPU server, there's $2 to a growing $3 of professional services around that, networking around that, storage around that.”
Clarke was speaking on a late Friday Q4 earnings call as Dell Infrastructure Solutions Group reported $9.3 billion in revenue for the quarter, up 10% on the previous quarter. Despite revenues overall being down, Dell’s share price soared on the results, which beat market expectations. The year “didn't go as planned, but I really like how we navigated it,” Clarke said.
The earnings of major hardware suppliers often furnish some unique insights into where they see momentum and how CIO priorities are evolving. Dell sells significant volumes to both cloud providers and enterprise customers, but to Clarke, it was the latter opportunity that was outsized. He said: “We believe the long-term AI action is on-prem where customers can keep their data and intellectual property safe and secure.
See also: Generative AI, data security, and toolchain maturity. An ecosystem evolves...
Pointing to a deck that suggested AI hardware and services represented a $152 billion opportunity by 2027, he said: “Growth is happening certainly in the public cloud, but increasingly more so in enterprises… 83% of all data is on-prem. We think AI moves to the data,” the Dell COO said.
He added: “More data will be created outside of the data centre going forward than inside the data centre today. That's going to happen at the edge of the network. A smart factory, an oil derrick or platform, a deep mine… We believe AI will ultimately get deployed next to where the data is created driven by latency… We've sold to financial services, business, engineering and consumer services companies. They're seeing vast deployments, proving out the technology. And some cases are using the tooling of the public cloud. And then they quickly find that they want to run AI on-prem because they want to control their data. They want to secure their data. It's their IP and they want to run domain-specific and process-specific models to get the outcomes they're looking for.”
Supply chains remain tight, he admitted: “Demand for the computational components to do AI exceeds the supply picture” – much of that due to incredible demand for GPUs and other hardware from suppliers like NVIDIA, with Clarke declining to share details, just saying “the backlog is a mixed array of [NVIDIA] H100s, H200s, H800s, as well as [AMD] MI300Xs, with varying supply commitments and varying delivery commitments.”
Dell’s revenue for the year was $88.4 billion, down 14% from fiscal year 2023. Net income was $3.1 billion, up an impressive 32%.