How Dell Storage Powers AI Factories

Key takeaways:

• Storage is the starting point for AI factories.
  Most enterprise data already lives on‑prem or at the edge, so storage design is the first decision for AI, not an afterthought.
• Dell file and object storage fuels AI workloads.
  Dell PowerScale, Dell ObjectScale and the Dell AI Data Platform form the storage layer that feeds GPUs and underpins the Dell AI Factory.
• You can get real AI outcomes without hyperscale builds.
  With dozens of GPUs, existing data centers and modernized storage, Dell and customers like Northwestern Medicine and the City of Amarillo are already seeing measurable impact.

AI factories live or die on data. That reality makes storage a strategic choice, not background plumbing.

In a recent episode of theCUBE’s AI Factory Series from the New York Stock Exchange, host John Furrier spoke with Kyle Leciejewski, Senior Vice President for North America at Dell. They discussed how enterprises are building AI factories on top of data and storage foundations they already trust.

Why data and storage sit at the center of AI

The AI models that create the most value run on a company’s unique proprietary data. That data must stay sovereign, secure and under tight control, especially when it supports high stakes decisions or differentiated services.

Most of that data already stays close to home. Kyle notes that about 83 percent of the world’s data sits in on‑premises data centers. He also points out that around half of new data now comes from the edge in the real world. As a result, storage becomes the first design question, not an afterthought. Enterprises decide how to store and protect data, then how to make it available for AI across core, cloud and edge.

Cost reinforces this focus. Kyle highlights that running AI on‑prem can be 70 to 75 percent more cost effective. The way teams architect storage and place data has a direct impact on that outcome.

From AI factory vision to storage portfolio

In the interview Kyle explains that AI factories start as an infrastructure conversation. Customers move from compute to networking, then quickly ask a storage question that sounds like “how do I feed these hungry GPUs” with file, object and unstructured data. After that they move into thermals, cabling and how everything is racked and integrated. The goal is a turnkey AI factory that feels like the easy button.

In Dell’s broader storage portfolio platforms such as Dell PowerScale for scale‑out file and Dell ObjectScale for object storage deliver those AI‑ready file and object capabilities. Together they form the storage foundation of the Dell AI Data Platform. That platform underpins the Dell AI Factory and moves, manages and protects data across the AI lifecycle. This is the context for the AI factory discussions Kyle is having with customers in the field.

Designing storage for hybrid AI and the edge

For Kyle the old debate about public or private cloud is over. In practice it is hybrid. Crown jewel data lives in the data center. Cloud and emerging Neoclouds offer GPUs and services by API. At the same time new data keeps exploding at the edge.

From a storage point of view that often means it is easier to bring compute to the data than to drag massive datasets back to a central site. To support this, Dell extends its portfolio beyond the core into AI enabled PCs that run small language models at the edge. It also offers compact NVIDIA enabled systems like the GB10 that can support multi‑billion parameter models near where data is created. In addition, software such as NativeEdge helps automate lifecycle tasks across this distributed estate. Underneath all of this sits storage that keeps data close, feeds GPUs efficiently and keeps information protected as it moves.

Real AI outcomes built on data

Kyle ties these storage and data decisions to clear outcomes. Inside Dell, an AI code assistant now supports about 42 percent of the company’s code. A sales chatbot called Dell Sales Chat saves sellers three to five hours a week by curating content from across a distributed data estate. In services, insights from 250 million assets that dial in to Dell help teams identify the next best action in support.

Customer examples bring the point home. At Northwestern Medicine, Dell infrastructure supports radiology teams who deliver higher quality digital reads at higher volume with better outcomes. In the City of Amarillo a digital human that speaks more than 50 languages helps citizens find evacuation routes and emergency information. In each case storage and data design quietly enable better AI experiences.

What this inflection point means for storage leaders

Kyle calls this the biggest inflection point in more than 40 years of Dell’s history. He also notes that customers do not need planetary scale data centers or tens of thousands of GPUs to get started. In Dell’s own work many early wins come from dozens of GPUs, air cooled systems and existing data centers, backed by storage that teams have modernized for consolidation and efficiency.

The message for storage teams is direct. Decisions about where data lives, how it is stored and how fast it can move will play a major role in how an organization participates in AI. Storage is no longer just supporting the AI factory, it is helping shape it. To hear Kyle’s full perspective from the floor of the New York Stock Exchange, you can watch the complete AI Factory Series conversation on theCUBE.