Nvidia introduces RTX Pro servers, based on the recently launched RTX Pro GPUs. The servers support AI workloads and more.
At Computex, Nvidia introduces RTX Pro servers. These servers are equipped with the RTX Pro 6000 Blackwell Server Edition GPU and are part of a broader validation by Nvidia. These server solutions are designed to support various workloads, ranging from AI inference to graphic applications and scientific simulations.
They are based on the RTX Pro chips. These are more classic GPUs built on the previously launched Blackwell architecture. The broader lineup is developed with workstations in mind, but the top component RTX Pro 6000 is an outlier. This GPU exists in a server variant and sits somewhere between a powerful workstation GPU and a Blackwell-based AI chip like the Nvidia B200. The new reference designs standardize servers built with that RTX Pro 6000 Blackwell Server Edition.
read also
Nvidia RTX Pro Blackwell for Workstations: Up to 600 Watts, 96 GB GDDR7, and Confusing Variants
Ecosystem
Nvidia wants customers to combine RTX Pro servers with other Nvidia technologies such as Spectrum-X Ethernet, BlueField DPUs, and AI Enterprise software. Partners like Cisco, Dell, HPE, and Lenovo will offer these complete solutions.
Nvidia wants to convince the world that so-called ‘AI factories’ are a thing. With this self-coined marketing term, the company has been referring to powerful servers optimized for AI workloads for quite some time. Nvidia eagerly shares recommended hardware configurations and software stacks to help companies implement these AI factories, which neatly make use of the entire Nvidia ecosystem and leave no room for hardware from competitors.
In terms of software, Nvidia’s solutions enjoy broad support. This also applies to these RTX Pro servers. They are compatible with third-party software such as Ansys, Cadence, Red Hat, and Siemens.