Dell announced the Dell AI Factory during GTC 2024. Now, two years later, it has 4,000 customers. According to the company, early adopters are seeing a 2.6x return on their investment. With the new offering based on Nvidia Vera Rubin, it aims to make as many organizations as possible “AI-ready.”
These days, Dell is flipping the script on the ROI question. What happens if you don’t adopt AI? Granted, adoption is perfectly possible without proprietary hardware, but Dell is trying to win over new potential customers by highlighting three critical requirements the company aims to meet.
The three lessons
Dell has established these requirements based on its 4,000 customers. First and foremost, business data must be ready for AI. That seems like an impossible mission for most organizations to accomplish on their own. That’s why Dell is introducing a Data Orchestration Engine, an “intelligence layer” to prepare data of all shapes and sizes for AI use. Gradually, this engine should even become smarter and better adapt to your organization’s business processes.
In a broader sense, AI hardware needs to be more widely available, Dell argues. In other words: AI infrastructure “from desktop to data center.” It not only sounds good, but it also gradually relieves the burden on overloaded data center infrastructure. For example, Dell offers the Pro Max desktop with the GB10 or GB300 Grace Blackwell Ultra Desktop Superchip, as well as powerful servers based on the new Nvidia Rubin. More on that later.
The third lesson is that deployment has simply been too slow for many. A modular architecture combined with the Dell Automation Platform should significantly shorten the deployment timeline. The company’s own Agentic AI Platform, developed in collaboration with Cohere North and DataRobot, also offers ready-to-use applications for organizations that are not yet able to handle the orchestration of agents themselves. Those who need more help can turn to Dell Accelerator Services, the company’s consulting and training division.
Data Center Level
Naturally, Dell is also updating its data center products to accompany the launch of Nvidia Vera Rubin. The PowerEdge XE9812 is the liquid-cooled rack based on the Nvidia Vera Rubin NVL72. Other rack-mounted options featuring the Nvidia HGX Rubin NVL8 include the PowerEdge XE9880L, XE9882L, and XE9885L.
The options listed above are primarily GPU-driven servers, even though Nvidia’s Vera CPU is a key component for performance. More common variants of Dell servers with Nvidia chips use “regular” x86 chips, such as the PowerEdge R770, R7715, and R7725. The Vera CPU is, however, included in the PowerEdge R9822 and M9822.
In addition, Dell offers enhanced networking options with the PowerSwitch SN6000 series, which features liquid-cooled Nvidia Spectrum-6 Ethernet switches. The InfiniBand variant is the Q3300-LD and is also liquid-cooled.
Modular Growth
It is clear that Dell has learned from two years of AI Factory. Now the next step is to make the technology more accessible and modular. Cleaning up data to be “AI-ready” increasingly seems like a task for hardware and software players rather than the customer.
This means that OEMs like Dell can claim a larger role in bringing AI to the market, where profitability is no longer the central focus. According to Dell, the reason is crystal clear: organizations will notice it sooner if they don’t jump on the AI bandwagon.
Read also: Europe’s data center market booms: is it ready to take on the US?