Dell Technologies expands its AI offering with new infrastructure solutions, software, and services built around Nvidia hardware.
Dell launches new servers tailored for AI workloads. The devices are built around accelerators from Nvidia, and fit within the AI factory narrative that Nvidia CEO Jensen Huang has been selling for almost two years. Behind this term are powerful servers, optimized for AI. Dell eagerly plays for team Nvidia and adopts the marketing term in the form of the Dell AI Factory with Nvidia.
New Servers
The latest additions to that portfolio include servers, storage solutions, and network equipment. Dell introduces, among others, air- and liquid-cooled PowerEdge XE9780 and XE9785 servers, which support up to 192 NVIDIA Blackwell Ultra GPUs.
In combination with Dell IR7000 racks, these can be expanded to 256 GPUs. These systems build upon the previous generation PowerEdge XE9680 and offer, according to Dell, up to four times faster training of large language models.
read also
Dell Refreshes Storage Portfolio: Focus on Flash and Efficiency with Power
The PowerEdge XE9712, in turn, combines NVIDIA GB300 NVL72 with Dell PowerCool technology. This system is optimized for inference tasks at rack scale. Also, starting in July, a version of the PowerEdge XE7745 server will be available that is suitable for applications such as robotics and digital twins, supporting up to eight GPUs.
Additionally, Dell expands its network portfolio with new switches, including the PowerSwitch SN5600 and NVIDIA Quantum-X800 InfiniBand. These are intended to facilitate data transfer within AI infrastructure. This is important: GPU servers are expensive devices. To utilize them optimally, the GPUs must remain saturated with data.
Software Updates
Besides hardware, Dell also introduces updates for its AI data platform. Dell ObjectScale becomes more compact and supports higher performance with lower latency. The integration of NVIDIA’s NIXL libraries and KV cache makes the platform more suitable for large-scale inference workloads. ObjectScale can now also handle S3 via RDMA, which, according to Dell, yields significant improvements in throughput and CPU load.
The Dell AI Factory with NVIDIA further offers software integrations such as NVIDIA AI Enterprise, NeMo microservices, and Red Hat OpenShift. Dell also provides new managed services that help companies manage the entire AI stack, including infrastructure and software. These services include 24/7 monitoring and maintenance.