A complete, factory-integrated NVIDIA Vera Rubin NVL72 rack with 72 Rubin GPUs and 36 Vera CPUs in a single NVLink 6 domain, sold and supported by Cisco and Supermicro and managed through Cisco Nexus One, delivered without a coolant distribution unit and paired with in-row CDU infrastructure scoped through Cisco sales.

This is the Rubin-generation successor to the GB300 NVL72 rack in the Cisco AI compute portfolio, ordered as one product. The underlying hardware is the Supermicro NVIDIA Vera Rubin NVL72 system, built on the third-generation NVIDIA MGX rack architecture: 18 liquid-cooled compute trays, 9 NVLink switch trays, and 4 power shelves, integrated and tested as a complete rack before delivery. The rack ships without a coolant distribution unit; it is designed for in-row CDU deployments, and all liquid cooling infrastructure is scoped separately. For information on any liquid cooling components, or for a quote on liquid cooling products specifically, contact your Cisco sales specialist. Cisco handles the selling motion, support through Cisco CX Services, and lifecycle management through Cisco software. Supermicro builds and integrates the hardware.

Each compute tray carries four NVIDIA Rubin GPUs and two NVIDIA Vera CPUs (88 custom Arm cores each), the rack totals 72 Rubin GPUs and 36 Vera CPUs connected through sixth-generation NVLink at 3.6 TB/s per GPU, 260 TB/s of aggregate scale-up bandwidth. GPU memory totals 20.7 TB of HBM4; with 54 TB of CPU-attached LPDDR5X, the rack presents roughly 75 TB of fast memory. Rack-level compute reaches up to 3.6 exaflops of NVFP4 inference performance. Target workloads are frontier-scale training and agentic and reasoning-model inference at the highest density NVIDIA offers.

Table 1 summarizes the primary features and benefits of the platform

Table 1. Features and benefits

The generational leap, same rack discipline: NVLink 6, HBM4, and 75 TB of fast memory.

Vera Rubin NVL72 is what the rack-as-computer idea looks like one generation later. Against the GB300 NVL72 it replaces, the same 72-GPU, 36-CPU footprint moves to NVLink 6 at twice the per-GPU bandwidth, HBM4 at roughly 1.4 PB/s of aggregate memory bandwidth, a purpose-built Vera CPU instead of Grace, ConnectX-9 scale-out at up to 1.6 Tb/s per GPU, and BlueField-4 DPUs. Total fast memory roughly doubles to 75 TB. Nothing about the operating model changes: one Cisco order, dedicated Cisco and Supermicro support contracts, one management plane (for Saas deployments) through Cisco Nexus One (Cisco Hyperfabric).

The cooling boundary is explicit. This rack is offered without a coolant distribution unit and is designed for in-row CDU infrastructure that serves multiple racks. It can also be deployed with Liquid-to-Air Sidecar CDUs. For information on any liquid cooling components, or for a quote on liquid cooling products specifically, contact your Cisco sales specialist.

Cisco offers this Supermicro platform in a defined rack configuration, listed in Table 2. Final specifications are yet to be defined.

Table 2. Cisco selected configuration, NVIDIA Vera Rubin NVL72 rack

Information about Cisco’s Environmental, Social and Governance (ESG) initiatives and performance is provided in Cisco’s CSR and sustainability reporting.

Table 3. Cisco Environmental Sustainability Information

Table 4 reflects NVIDIA platform specifications and Supermicro announcements for the Vera Rubin NVL72 rack. Supermicro has not yet published the full data sheet for this SKU; product specifications may change without notice.

Table 4. Specifications, Supermicro NVIDIA Vera Rubin NVL72 rack

Table 5 lists what must be in place before installation. This is a rack-scale liquid-cooled system delivered without a CDU; for information on any liquid cooling components, or for a quote on liquid cooling products, contact your Cisco sales specialist.

Table 5. System requirements

The rack is a Supermicro-built system offered through Cisco under the part number below, orderability starting October 2026. No coolant distribution unit is included. Liquid cooling components (in-row CDUs, manifolds, and related infrastructure) are quoted separately: contact your Cisco sales specialist for information on any liquid cooling components or for a quote on liquid cooling products specifically. To order, contact your Cisco account representative or visit the Cisco Ordering Home Page.

Table 6. Ordering information

Warranty terms for this Cisco-offered Supermicro platform are standard Supermicro warranty terms. Support is delivered through Cisco CX Services and Supermicro under the applicable service contracts, contact your Cisco sales specialist for more details.

Cisco and Partner Services

Cisco CX and Supermicro Services cover this platform from planning through deployment and ongoing operations, with dedicated support contracts. For more information, visit https://www.cisco.com/go/services.

Cisco Capital

Flexible payment solutions to help you achieve your objectives

Cisco Capital makes it easier to get the right technology to achieve your objectives, enable business transformation and help you stay competitive. We can help you reduce the total cost of ownership, conserve capital, and accelerate growth. In more than 100 countries, our flexible payment solutions can help you acquire hardware, software, services and complementary third-party equipment in easy, predictable payments. Learn more.

Learn more

For more on Cisco AI infrastructure, visit https://www.cisco.com/go/ai. Full platform specifications will be available on the Supermicro Vera Rubin solutions page once published.

Learn more about Cisco and Supermicro partnership:
https://www.cisco.com/site/us/en/solutions/global-partners/supermicro/index.html.