Dell Technologies
The largest enterprise channel for NVIDIA compute. Dell engineers the PowerEdge XE series and IR7000 rack infrastructure itself, and wraps it in the supply chain, services and financing a regulated enterprise expects. Lead times track demand from its largest AI customers.
- Integrated Rack 7000 (IR7000): own 21-inch ORv3-based rack platform purpose-built for liquid cooling, up to 480kW per rack, near 100% heat capture
- Own direct-to-chip liquid cooling design with cold plates, integrated busbars and DLC manifolds built into the IR7000
- iDRAC: in-house embedded management controller with silicon root of trust; factory OpenBMC option
- Custom 1RU compute sleds engineered for the XE9712 GB200 NVL72 rack-scale system
- GPU baseboards largely follow NVIDIA reference designs; Dell's original engineering is at the rack, cooling, sled and management layers
| Model | Platform | Form factor | Cooling |
|---|---|---|---|
| PowerEdge XE9712 First shipped to CoreWeave | GB200 NVL72 | rack-scale (1RU sleds in IR7000) | DLC |
| PowerEdge XE8712 2 Grace CPUs plus 4 B200 GPUs per node, up to 264kW per rack | GB200 NVL4 | up to 36 nodes / 144 GPUs per IR7000 rack | DLC |
| PowerEdge XE9680 8-GPU HGX baseboard, six PSUs totaling 19,200W | HGX H100/H200/B200 | 10U | air |
| PowerEdge XE9680L Liquid-cooled variant for B200's higher power | HGX B200 | 10U | DLC |
| PowerEdge XE9812 (PowerRack) First Vera Rubin rack shipped, to CoreWeave | Vera Rubin NVL72 | rack-scale | DLC |
| PowerEdge XE7745 RTX PRO 6000 Blackwell Server Edition, positioned for inference | PCIe | 4U | air |
- Dell AI Factory with NVIDIA: named joint program and portfolio brand
- System builder for HGX (XE9680/XE9680L), GB200 NVL72 rack-scale (XE9712) and GB300 NVL72
- Named in NVIDIA's Vera Rubin builder lists; claims first to ship Vera Rubin rack systems (XE9812 delivered to CoreWeave)
- XE-series servers carry NVIDIA-Certified Systems designation
- Austin, Texas area (server assembly near Round Rock HQ)
- Memphis, Tennessee (rack assembly reported for the xAI Colossus buildout)
- Leading AI server market share, ~20% in 2024 per ABI Research
- Over $64B cumulative AI server orders; $51.3B AI backlog after Q1 FY2027; FY2027 AI server revenue guidance ~$60B
- Marquee customers: CoreWeave (IR7000, XE9712, first Vera Rubin systems) and xAI (~$5B reported GB200 deal)
- Management states component demand outpaces supply, extending lead times
Is Dell Technologies an OEM, ODM or integrator for NVIDIA AI servers?
Dell Technologies sits in the "Global OEMs" tier of our index. Integrated Rack 7000 (IR7000): own 21-inch ORv3-based rack platform purpose-built for liquid cooling, up to 480kW per rack, near 100% heat capture
Does Dell Technologies design its NVIDIA servers in-house?
The largest enterprise channel for NVIDIA compute. Dell engineers the PowerEdge XE series and IR7000 rack infrastructure itself, and wraps it in the supply chain, services and financing a regulated enterprise expects. Lead times track demand from its largest AI customers.
Which NVIDIA platforms does Dell Technologies ship?
Dell Technologies ships systems on GB200 NVL72, GB200 NVL4, HGX H100/H200/B200, HGX B200, Vera Rubin NVL72, PCIe. The systems table on this page lists the specific models.
Where does Dell Technologies manufacture its AI servers?
Austin, Texas area (server assembly near Round Rock HQ). Memphis, Tennessee (rack assembly reported for the xAI Colossus buildout)