Lenovo NVIDIA GB300 NVL72 Rack Scale AI

The Lenovo NVIDIA GB300 NVL72 is a rack scale solution built on the NVIDIA Blackwell architecture, which delivers groundbreaking advancements in accelerating computing, leading the new era of AI with optimized compute and increased memory.

The GB300 NVL72 is the NVIDIA flagship rack-scale AI offering. It is an end-to-end NVIDIA reference design adopted by the market.

The NVIDIA GB300 NVL72 is a platform designed for AI reasoning performance and efficiency, a scalable building block approach design, featuring 72 NVIDIA Blackwell Ultra GPUs and 36 Arm-based NVIDIA Grace CPUs in a hybrid cooling standard rack.

The GB300 NVL72 uses a hybrid cooling architecture: CPUs, GPUs, and NVSwitch components are liquid‑cooled, while OSFP modules, storage drives, and PDB components are air‑cooled.

The components of the AI rack solution are designed to be compliant with the NVIDIA MGX rack. The liquid cooling solution consists of a CDU, rear manifold, quick disconnects, and cold plates for the CPUs, GPUs, ConnectX-8 network adapters, and all NVSwitch components.

360° View Full 3D Tour

The GB300 NVL72’s integrated NVLink switch enables every GPU in the rack to communicate as part of one unified fabric. Each rack delivers 72 NVIDIA Blackwell GPUs interconnected through the 5th generation NVIDIA NVLink™ technology, forming a single high performance compute domain ideal for large‑scale multimodal reasoning, high‑performance inference and LLM training workloads.

The Lenovo NVIDIA GB300 NVL72 is a rack scale solution for multiple interconnected NVIDIA GPUs. The solution consists of 18 GPU Compute trays, 9 Switch trays and either 6 or 8 Power Shelves all connected via a rear cable cartridge backplane.

Turnkey AI Rack Design

Engineered for large-scale Artificial Intelligence (AI) and High Performance Computing (HPC), Lenovo NVIDIA GB300 NVL72 excels in intensive simulations and complex modeling. It is designed to handle modeling, training, simulation, rendering, financial tech, scientific research, technical computing, grid deployments, and analytics workloads in various fields such as research, life sciences, energy, engineering, and financial simulation.

The NVIDIA GB300 NVL72 by Lenovo features a mix of air and liquid cooling to meet its demanding power requirements. The NVIDIA GB300 NVL72 integrates seamlessly into a standard 19" rack cabinet. Each rack consumes 135 kW TDP; up to 155 kW peak depending on workload and EDP behavior. At the rack level, the split between air and liquid heat capture is about 10% to air and 90% to liquid.

This design ensures easy serviceability and extreme performance density, making the NVIDIA GB300 NVL72 the go-to choice for compute clusters of all sizes - from departmental/workgroup levels to the world’s most powerful supercomputers – from Exascale to Everyscale®.

Each rack delivers 72 NVIDIA Blackwell GPUs interconnected through the 5th generation NVIDIA NVLink™ technology, forming a single high performance compute domain ideal for large‑scale multimodal reasoning, high‑performance inference and LLM training workloads.

Direct water-cooled solutions are factory-integrated and are re-tested at the rack-level to ensure that a rack can be directly deployed at the customer site. This careful and consistent quality testing has been developed as a result of over a decade of experience designing and deploying DWC solutions to the very highest standards.

Scalability and performance

The Lenovo NVIDIA GB300 NVL72 is based on a GB300 rack infrastructure which offers the following features to boost performance and improve scalability:

• The NVIDIA GB300 NVL72 Rack Level GPU System
  • One rack with 1U Compute Trays, 72 GPUs with NVLINK with non‑scalable (NVLink domain is contained within a single rack; no inter‑rack NVLink scaling) NVSwitches.
  • Cooling solution is a hybrid of liquid and air-cooled components
  • Power shelves either 6 or 8. Quantity dependent on rack configuration and power redundancy requirement
• The 1U Compute Tray leverages the NVIDIA GB300 NVL72 architecture:
  • Two NVIDIA Grace processors
  • Four NVIDIA Blackwell SXM7 GPUs
  • Highspeed NVLink interconnects between GPUs
• Each compute tray has up to 2x PCIe Gen 5 x16 slots for high-speed networking, depending on the configuration.
• Each compute tray also supports up to 8x E1.S 3.84T drive bays, depending on the configuration, for internal storage and 1x 1.92T M.2 NVMe SSD for OS boot
• All drives are E1.S NVMe drives with PCIe Gen 5 x4 host interface, to maximize I/O performance in terms of throughput, bandwidth, and latency.
• The compute tray includes one 1Gb Ethernet onboard port for debug networking. One BMC card for system management (Out Of Band port supports for BMC management communication). High speed host networking can be added through the included PCIe slots.
• The compute tray offers PCI Express 5.0 (PCIe Gen 5) I/O expansion capabilities. A PCIe Gen 5 x16 slot provides 128 GB/s bandwidth, enough to support 800GbE via OSFP network connections (1x 1pt NVIDIA BlueField-3 adapter). 400/200GbE supported via breakout DAC/AOC cables. Two Gen 5 x16 connections (32 lanes total) provide the bandwidth needed for an 800 GbE connection using the ConnectX-8 8180 network adapter (2x 2port ConnectX-8 adapters).
• The 1U NVLink™ Switch Tray leverages the NVIDIA GB300 NVL72 architecture:
  • One GB300 NVL72 NS Switch Tray Base (non‑scalable)
  • One GB200 NVL72 Reduntant Power Shelf Solution
• The NVIDIA SN2201 Management Switch leverages the NVIDIA GB300 NVL72 architecture:
  • One NVIDIA SN2201 48 port DC switch
  • One GB200 NVL72 Redundant Power Shelf Solution

Energy efficiency

The direct water cooled solution offers the following energy efficiency features to save energy, reduce operational costs, increase energy availability, and contribute to a green environment:

• Water cooling eliminates power that is drawn by cooling fans in the enclosure and dramatically reduces the required air movement in the server room, which also saves power in the data center due to the reduced need for air conditioning.
• Water chillers may not be required with a direct water cooled solution. Chillers are a major expense for most geographies and can be reduced or even eliminated because the water temperature can now be 10°C to 35°C in an air-cooled environment.
• Heat energy absorbed may be reused for heating buildings in the winter, or generating cold through Adsorption Chillers, for further operating expense savings.
• The processors and other microelectronics are run at lower temperatures because they are water cooled, which uses less power, and allows for higher performance.
• The power solution consists of 50V power shelves delivering power to the bus bar and complying with the standard MGX v1.1 rack specifications.

Manageability and security

The following powerful systems management features simplify local and remote management of the NVIDIA GB300 NVL72 system:

• Support for industry standard management protocols, IPMI 2.0, Redfish REST API, serial console via IPMI
• The NVIDIA GB300 NVL72 is enabled with Lenovo HPC & AI Software Stack, so, you can support multiple users and scale within a single cluster environment.
• Lenovo HPC & AI Software Stack provides our HPC customers you with a fully tested and supported open-source software stack to enable your administrators and users with for the most effective and environmentally sustainable consumption of Lenovo supercomputing capabilities.
• Our Confluent management system provides an interface designed to abstract the users from the complexity of HPC cluster orchestration and AI workloads management, making open-source HPC software consumable for every customer.
• Integrated Trusted Platform Module (TPM) 2.0 support enables advanced cryptographic functionality, such as digital signatures and remote attestation.
• Supports Secure Boot to ensure only a digitally signed operating system can be used.
• Industry-standard Advanced Encryption Standard (AES) NI support for faster, stronger encryption.
• For management: BMC Server Management chip (ASPEED AST2600) and HMC.
• Power shelves are located at top and bottom of the rack (33kW each).
• The Baseboard Management Controller (BMC) in the compute trays is used in conjunction with the Host Module Controller (HMC) to provide Out-of-band management.

Availability and serviceability

The GB300 Compute tray and GB300 NVL Switch Tray provide the following features to simplify serviceability and increase system uptime:

• Designed to run 24 hours a day, 7 days a week
• With hybrid cooling, water cooling and air cooled systems, some fans are required.
• Power Shelves to minimize downtime as current is shared between multiple shelves.
• Toolless cover removal on the nodes provides easy access to upgrades and serviceable parts such as adapters and drives.
• The system desired to support non-disruptive firmware upgrade for any programmable.
• FRUs (Field Replaceable Unit) that are supported by the design. FRUs are parts that can be field replaceable.
• Any cabling connectors desired to include a latch and a fool proof design to prevent users from plugging it in backwards.
• The external debug console port desired to have terminal access to each of the major components (CPU/OS, BMC) for debug and diagnostic purposes.
• There is a three-year customer replaceable unit and onsite limited warranty, with next business day 9x5 coverage. Optional warranty upgrades and extensions are available.

The table below shows the main system components of the rack scale solution. This is to depict an overall hardware configurability, and capability, for the complex set of feature options that the system needs to be able to support in hardware. Not all configurable combinations, or feature options, are shown here; such as DIMM memory options, TPM, or storage options as examples. Also not shown are the TOR switches which are dependent on customer network topology configuration.

The GB300 Compute trays, GB300 NVL Switch Trays, Power Shelves and Top of Rack switches are all integrated and rack mounted. Below is a high-level rack overview of the NVL72 rack configuration.

Figure 2. Front view of the Lenovo NVIDIA GB300 NVL72 rack

The front of the compute server is shown in the following figure.

Figure 3. Front view of the GB300 Compute Tray

The following figure shows the main components on the rear of the Compute Tray configuration that includes Quick Disconnect (QD) ports and Busbar clip.

Figure 4. Rear view of the GB300 Compute Tray

The following figure shows key components internal to the Compute Tray.

Figure 5. Inside view of the GB300 Compute in the water-cooled tray

The NVIDIA GB300 NVL72 is offered with the following integrated systems:

• GB300 Compute tray
• GB300 NVL Switch Tray
• SN2201 Management Switch
• Power Shelf

The following figure shows the architectural block diagram of the GB300 Compute Tray. This configuration includes 4x GPUs and 3x PCIe Gen5 x16 slots all directly connected to the CPUs. This configuration optimizes network connectivity while maximizing computational efficiency, ensuring cost-effectiveness in performance.

Figure 7. 1U GB300 Compute Tray system architectural block diagram

The following figure shows the architectural block diagram of the GB300 NVL Switch Tray. This switch interfaces by connecting the Compute Tray nodes, GPUs, 800Gb/s OSFP network interfaces and drives all together using ConnectX-8 switches. The AI Training configuration is optimized for scale-out GPU performance which provides advantages in large AI training workloads.

Figure 8. GB300 NVL Switch Tray system architectural block diagram

For more information about the two configurations, see the Configurations section.

The following table lists the standard specifications of the Lenovo NVIDIA GB300 Compute tray.

Notes:

Onsite spares (compute nodes) would be needed to support an onsite Next Business Day Service Level target. 2% of the order recommended quantity. For Compute Tray base alternatives available for selection, see the Spare Compute Trays section.

The following table lists the standard specifications of the Lenovo GB300 NVL Switch Tray.

The NVIDIA GB300 NVL72 rack is configured by using the configure-to-order (CTO) process with the Lenovo Cluster Solutions configurator (x-config) or in Lenovo Data Center Solution Configurator (DCSC).

The following table lists the base CTO models and base feature codes

The Lenovo NVIDIA GB300 NVL72 node is equipped with high-performance GPUs. It is RoHS complaint and meets all the environmental certifications.

There are two primary configurations offered with the NVIDIA GB300 NVL72. Block diagrams for Compute Tray and Switch Tray leveraged in these configurations are shown in the System architecture section.

The following table lists the Bill of Materials for the Non-redundant Power with 6 Power Shelves

The following table lists the Bill of Materials for Redundant Power with 8 Power Shelves

For more information about the, see the Lenovo NVIDIA GB300 NVL72 datasheet.

The following table lists the Configurators used to create the above configurations.

Configurator tips:

• The following Compute Tray Base alternatives are available for selection: 
  • For Rack Installed compute Trays: Select Lenovo Feature Code CABF - 1U 4GPU GB300 DWC Front IO
  • For Spare compute Trays with individual node packaging: Select Lenovo Feature Code CD96 - 1U 4GPU GB300 DWC Front IO w Package  - Mainly for customer who would like to order for replacement or as a spare.
  • For Service Spare compute Trays with individual node packaging: Select Lenovo Feature Code CCQD - 1U 4GPU GB300 DWC Front IO w/o E1.S, w Package" – For Service, for repair purpose.

To guarantee a consistent next business day response objective and minimize potential downtime, we strongly recommend the purchase of on-site spare Compute Trays. Lenovo recommends a minimum spare-to-production ratio of 2%, however the ideal number of spares depends on the scale of your deployment. Contact Lenovo Services for tailored guidance on the appropriate number of spares.

The following Compute Tray Base alternatives are available for selection:

The NVIDIA GB300 NVL72 leverages two NVIDIA GB300 NVL72 host processor motherboards (HPMs) per tray that integrate processors, memory and GPU accelerators in a powerful package.

• Processors
• Memory
• GPU accelerators

Each of the NVIDIA GB300 NVL72 HPMs include one NVIDIA Grace processor, as part of the NVIDIA GB300 NVL72 module. The processors are standard in all configurations of the NVIDIA GB300 NVL72.

Each processor has the following features:

• 72 Arm Neoverse V2 cores with 4x 128b SVE2
• 3.1 GHz base frequency
• 3.0 GHz all-core SIMD frequency
• L1 cache: 64KB i-cache + 64KB d-cache
• L2 cache: 1MB per core
• L3 cache: 114MB
• 64 PCIe Gen5 lanes (implemented as 8x x8)
• 300 GB/s bi-directional NVLink-C2C connectivity between processors
• 450 GB/s bi-directional NVLink connectivity to two GPUs each
• NVIDIA Scalable Coherency Fabric (SCF) with 3.2 terabytes/s bisection bandwidth interconnecting Cores, Memory and IO
• 300W TDP (includes power for memory)

The NVIDIA Grace processor is integrated using Ball Grid Array (BGA) technology and is soldered onto the NVIDIA GB300 NVL72 HPM board. To replace the CPU, it is necessary to replace the entire module.

The NVIDIA GB300 NVL72 HPM boards incorporate processor memory via LPDDR5x modules.

Features of the memory subsystem are as follows:

• 960GB of LPDDR5X ECC memory in a fixed configuration, 480GB per processor
• Implemented using 8x 128GB modules (4 per CPU)
• Peak memory bandwidth per CPU: Up to 384 GB/s
• Memory clock per CPU: 4266MT/s

The Single Error Correction (SEC) and Double Error Detection (DED) algorithms as part of the ECC are designed to mitigate multi-bit aliasing and correct errors effectively. Furthermore, features such as parity, scrubbing, poisoning, and page off-lining are implemented to enhance Reliability, Availability, and Serviceability (RAS).

The NVIDIA GB300 NVL72 HPM boards support four onboard NVIDIA B300 GPUs. The following table lists the specifications for each GPU and for the full package. Measured performance will vary by application.

The following table lists the specifications for each GPU.

The following figure shows how the GPUs interconnect.

Figure 9. Compute Tray NVLink topology

The NVIDIA GB300 NVL72 node supports the following drives.

• Up to 8x E1.S drives for data storage.
  • These are front accessible.
• 1x M.2 NVMe drive for OS boot
  • This is an internal drive - not front accessible and non-hot-swap.

The following images shows drives that come standard in the Compute Tray and their location.

Figure 10. NVIDIA GB300 NVL72 internal drive bays

Each of the 8 drive bays supports either of the following:

• E1.S 7.68TB drives
• E1.S 3.84 TB drives

Configuration notes:

• E1.S installation rule:
  • 1. Default: 4 x E1.S SSDs, install bays 0, 2, 4, 6
  • 2. If select 8 x E1.S SSDs, install bays 0, 1, 2, 3, 4, 5, 6, 7
  • Dummy Panels required to fill in unused bay slots
• The Compute Tray supports E1.S NVMe SSDs for data and 1× M.2 NVMe SSD for OS boot
• The drives are connected to onboard controllers; RAID functionality is provided using the operating system features
• Boot drive is the same as other drives but use switched PCIe and may have lower performance. The M.2 riser kit (feature C58W) is needed. The M.2 riser kit is not field upgradable.
• The same PCIe lanes can be used either for PCIe adapters or for NVMe drives.
• NVMe drives are connected to CPUs as follows:
  • Drives 0-3 are data storage drives connected to CX8 / CPU 1
  • Drives 4-7 are data storage drives connected to CX8 / CPU 2

The feature codes to select the appropriate storage cage are listed in the following table.

The drives of the NVIDIA GB300 NVL72 are connected to integrated NVMe storage controllers.

If desired, RAID functionality is provided by the installed operating system (software RAID only)

The following table lists the drive options for internal storage of the Compute Tray or server.

The NVIDIA GB300 NVL72 node supports up to 3x front-accessible slots, depending on the configuration:

• Slots 1, 2: PCIe 5.0 x16 slots connected to CPU 1
• Slots 3: PCIe 5.0 x16 slot connected to CPU 2

The location of the slots is shown in the following figure.

Figure 11. NVIDIA GB300 NVL72 PCIe slots

Slot availability is based on the configuration selected. Derived parts come standard within the  7DLZCTO2WW GB300 Compute tray BOM.

PCIe adapters and data drives share the same PCIe connections to the processors. The following table lists the supported data and boot drives.

OSFP 800Gb ports

The GB300 Compute tray configuration includes two OSFP interface boards (feature CCCS), each providing two OSFP ports. Supported connections are each up to 800 Gb/s (a total of 3200 Gb/s with 4x ports).

For more information, including the supported transceivers and cables, see the adapter product guide:

• ConnectX-8: https://lenovopress.lenovo.com/lp2163-thinksystem-nvidia-connectx-8-8180-800gbs-osfp-pcie-gen6-x16-adapter

For more information, including the supported transceivers and cables, see the adapter product guide:

• ConnectX-8: https://lenovopress.lenovo.com/lp2163-thinksystem-nvidia-connectx-8-8180-800gbs-osfp-pcie-gen6-x16-adapter

The Lenovo NVIDIA GB300 NVL72 offering supports the two most common fluids used the de-ionized water and PG25.

The customer can choose the coolant they want to use:

• Deionized (DI) water (recommended); Rack ships without water
• PG25 (glycol-based)

One of the most notable features of the Lenovo NVIDIA GB300 NVL72 offering is direct water cooling. Direct water cooling (DWC) is achieved by circulating the cooling water directly through cold plates that contact the GPUs, CPU and memory in the Compute Trays and through cold plates in the NVL Switch Trays.

The NVIDIA GB300 NVL72 by Lenovo features a mix of air and liquid cooling to meet its demanding power requirements. Each rack consumes 135 kW TDP; up to 155 kW peak depending on workload and EDP behavior. At the rack level, the split between air and liquid heat capture is about 10% to air and 90% to liquid.

GPU Compute Tray contains a hybrid cooling solution. The CPU, GPU, and HBM (GPU memory) are liquid cooled while the front option components, Power Distribution Board, and M.2 are air cooled. Therefore, while the air cooled component temperatures drive fans speeds, the liquid cooled components do not change fluid flow as a function of component temperature.

There are several components’ temperatures monitored to properly drive the eight system fan pairs. They are OSFP transceivers, QSFP transceivers, BlueField-3, Power Distribution Board components, ambient air, BMC, HMC, M.2, and E1.S components.

One of the main advantages of direct water cooling is the water can be relatively warm and still be effective because water conducts heat much more effectively than air. Depending on the environmentals like water and air temperature, effectively 90% of the heat can be removed by water cooling; in configurations that stay slightly below that, the rest can be easily managed by a standard computer room air conditioner.

Allowable inlet temperatures for the water can be as high as 45°C (113°F) with the NVIDIA GB300 NVL72 for real-world applications. In most climates, water-side economizers can supply water at temperatures below 45°C for most of the year. This ability allows the data center chilled water system to be bypassed thus saving energy because the chiller is the most significant energy consumer in the data center. Typical economizer systems, such as dry-coolers, use only a fraction of the energy that is required by chillers, which produce 6-10°C (43-50°F) water. The facility energy savings are the largest component of the total energy savings that are realized when the NVIDIA GB300 NVL72 is deployed.

The advantages of the use of water cooling over air cooling result from water’s higher specific heat capacity, density, and thermal conductivity. These features allow water to transmit heat over greater distances with much less volumetric flow and reduced temperature difference as compared to air.

For cooling IT equipment, this heat transfer capability is its primary advantage. Water has a tremendously increased ability to transport heat away from its source to a secondary cooling surface, which allows for large, more optimally designed radiators or heat exchangers rather than small, inefficient fins that are mounted on or near a heat source, such as a CPU.

The NVIDIA GB300 NVL72 offering uses the benefits of water by distributing it directly to the highest heat generating node subsystem components. That energy savings results from the removal of the system fans and the lower operating temp of the direct water-cooled system components.

The direct energy savings at the enclosure level, combined with the potential for significant facility energy savings, makes the NVIDIA GB300 NVL72 an excellent choice for customers that are burdened by high energy costs or with a sustainability mandate.

The liquid cooling solution consists of a coolant distribution unit (CDU), rear manifold, Disconnect Quick Connect (QD)s, and cold plates for the CPUs, GPUs, CX-8 NICs and all NVL Switch components.

The following table lists the reference design cooling percentages.

Water connections to the MGX Rack are provided using hoses that connect directly to the coolant distribution unit (CDU), inlet and return, either an in-rack CDU or in-row CDU. As shown in the following figure, hose lengths required depend on the placement of the Compute Trays and GB300 NVL Switch Trays in the rack cabinet.

The midplane in the chassis routes the water via QuickConnects to each of the Compute Tray, and each of the NVL Switch Trays.

The GB300 offers two options for the rack level liquid cooling manifold: bottom feed and top feed. Both options are shown below.

Figure 12. Water connections to the MGX Rack cabinet

Water connections are made up of the following:

• Short hoses permanently attached to the rear of the rack cabinet
• Intermediate hoses, orderable as three different pairs of hoses, based on the location of the components in the rack.
• Connection to the CDU through additional hoses or valves on the data center water loop. Lenovo Datacenter Services can provide the end-to-end enablement for your water infrastructure needs.

For additional information, see the Cooling section.

To support the onsite setup for the direct water-cooled solution, a service Kit is available providing a flow meter, bleed hose, pressure gauge and vent valve.

The liquid cooling manifold information is listed in the following table.

For details on Lenovo Neptune direct water-cooling standards, see the following document:

Lenovo Neptune Direct Water-Cooling Standards https://lenovopress.lenovo.com/lp2018-lenovo-neptune-direct-water-cooling-standards

The Lenovo NVIDIA GB300 NVL72 supports up to 6 or 8 power shelves. The use of water-cooled power conversion stations enables an even greater amount of heat can be removed from the data center using water instead of air-conditioning.

The Power Shelves supply internal system power to a 50V busbar. This innovative design merges power conversion, rectification, and distribution into these power units, a departure from traditional setups that require separate units, including separate power supplies, resulting in best-in-class efficiency.

The following table lists specifications for Power Shelf.

The following table lists Bus Bar specifications.

The following table lists other rack components.

The following image shows the power shelf.

Figure 13. Power shelf

The NVIDIA management is the BMC board option which provides advanced control, monitoring, and alerting functions. The BMC is used in conjunction with the Host Module Controller (HMC) to provide OOB management. The HMC interposer board is within the GPU Compute Tray, which connects the HMC board and the BMC board to the GB300 HPM via cables.

The HMC and BMC work together to provide the out-of-band manageability for the tray.

Topics in this section:

• Local management
• NVIDIA AI Enterprise

Additionally, the following table lists the features of the BMC Server Management.

The following table lists the features of the SBIOS.

NVIDIA AI Enterprise

The NVIDIA GB300 NVL72 is designed for NVIDIA AI Enterprise, which is a comprehensive suite of artificial intelligence and data analytics software designed for optimized development and deployment in enterprise settings.

NVIDIA AI Enterprise includes workload and infrastructure management software known as Base Command Manager. This software provisions the AI environment, incorporating the components such as the Operating System, Kubernetes (K8S), GPU Operator, and Network Operator to manage the AI workloads.

Additionally, NVIDIA AI Enterprise provides access to ready-to-use open-sourced containers and frameworks from NVIDIA like NVIDIA NeMo, NVIDIA RAPIDS, NVIDIA TAO Toolkit, NVIDIA TensorRT and NVIDIA Triton Inference Server.

• NVIDIA NeMo is an end-to-end framework for building, customizing, and deploying enterprise-grade generative AI models; NeMo lets organizations easily customize pretrained foundation models from NVIDIA and select community models for domain-specific use cases.
• NVIDIA RAPIDS is an open-source suite of GPU-accelerated data science and AI libraries with APIs that match the most popular open-source data tools. It accelerates performance by orders of magnitude at scale across data pipelines.
• NVIDIA TAO Toolkit simplifies model creation, training, and optimization with TensorFlow and PyTorch and it enables creating custom, production-ready AI models by fine-tuning NVIDIA pretrained models and large training datasets.
• NVIDIA TensorRT, an SDK for high-performance deep learning inference, includes a deep learning inference optimizer and runtime that delivers low latency and high throughput for inference applications. TensorRT is built on the NVIDIA CUDA parallel programming model and enables you to optimize inference using techniques such as quantization, layer and tensor fusion, kernel tuning, and others on NVIDIA GPUs. https://developer.nvidia.com/tensorrt-getting-started
• NVIDIA TensorRT-LLM is an open-source library that accelerates and optimizes inference performance of the latest large language models (LLMs). TensorRT-LLM wraps TensorRT’s deep learning compiler and includes optimized kernels from FasterTransformer, pre- and post-processing, and multi-GPU and multi-node communication. https://developer.nvidia.com/tensorrt
• NVIDIA Triton Inference Server optimizes the deployment of AI models at scale and in production for both neural networks and tree-based models on GPUs.

It also provides full access to the NVIDIA NGC catalogue, a collection of tested enterprise software, services and tools supporting end-to-end AI and digital twin workflows and can be integrated with MLOps platforms such as ClearML, Domino Data Lab, Run:ai, UbiOps, and Weights & Biases.

Finally, NVIDIA AI Enterprise introduced NVIDIA Inference Microservices (NIM), a set of performance-optimized, portable microservices designed to accelerate and simplify the deployment of AI models. Those containerized GPU-accelerated pretrained, fine-tuned, and customized models are ideally suited to be self-hosted and deployed on the NVIDIA GB300 NVL72.

Topics in this section:

• Security features

Security features

The server offers the following electronic security features:

• Support for Platform Firmware Resiliency (PFR) hardware Root of Trust (RoT)
• Firmware signature processes compliant with FIPS and NIST requirements
• Administrator and power-on password
• Integrated Trusted Platform Module (TPM) supporting TPM 2.0

The server is NIST SP 800-147B compliant.

The following table lists the security options for the NVIDIA GB300 NVL72.

The NVIDIA GB300 NVL72 supports the following operating systems:

• Ubuntu 22.04 with NVIDIA optimized HWE kernel (kopt)

The Lenovo 48U MGX Rack has the following dimensions:

• Width: 600 mm (23.6 inches)
• Height: 2,294 mm (90.3 inches)
• Depth: 1,068 mm (42.05 inches) - Expands to 1,200 mm with extension frame
• Weight: 185 kg (407.8 lbs)

GB300 Compute trays are installed onto the rack. Each GB300 Compute tray has the following dimensions:

• Width: 438 mm (17.3 inches)
• Height: 43.6 mm (1.72 inches)
• Depth: 766 mm (29.9 inches) (799 mm, including the water connections at the rear of the server)
• Weight: 29 kg (63.93 lbs)

The GB300 NVL Switch Tray has the following overall physical dimensions, excluding components that extend outside the standard chassis, such as EIA flanges, front security bezel (if any), and power conversion station handles:

• Width: 438 mm (17.3 inches)
• Height: 43.6 mm (1.72 inches)
• Depth: 770 mm (30.3 inches)
• Weight: 17 kg (37.5 lbs)

The NVIDIA SN2201 48 port DC switch has the following dimensions:

• Width: 438 mm (17.3 inches)
• Height: 43.6 mm (1.72 inches)
• Depth: 781 mm (28.3 inches)
• Weight: 11.39 kg (25.1 lbs)

The Power Shelf has the following dimensions:

• Width: 448 mm (17.6 inches)
• Height: 43.6 mm (1.72 inches)
• Depth: 718.5 mm (28.3 inches)

The following table lists the detailed dimensions. See the figure below for the definition of each dimension.

Figure 15. Enclosure dimensions

The GB300 Compute tray or server trays and GB300 NVL Switch Trays are supported in the following environment.

Topics in this section:

• Air temperature and humidity
• Water requirements
• Particulate contamination

Air temperature and humidity

Air temperature/humidity requirements:

• ASHRAE standard A2
  • Temperature operating:
    • Meets ASHRAE standard A2 (10°C to 35°C)
    • Maximum rate of change (°C /hr) ≤ 20
    • Operating temperature de-rated 1C per 300m (10,00ft) to 1,371m (4,500ft) above sea level
  • Humidity operating:
    • 20% - 80%, non-condensing
    • Humidity transition rate < 10%/hr
• Acoustics idle/working mode:
  • Fan control from BMC with multiple temperature input
  • Sound pressure level (LpAm): up to 82.9 dB
  • Estimated that the rack will operate around 92dBA from rack rear, typical ambient and rack exercised
• Altitude:
  • Supports 0 – 5,000 ft (0 - 1,524m)
• Temperature Non-operating: w/o PKG
  • -40C to 70C (Rack shipped with pressurized nitrogen)
  • Non-operating with Di-H20: 5-70C
  • Non-operating with PG25: 10-70C

Water requirements

For details on the NVIDIA GB300 NVL72 direct water-cooling standards, see the following table:

Standard specifications of the Lenovo GB300 NVL Switch Tray.

• Operating water temperature: 2°C to 50°C (35.6°F to 122°F) (ASHRAE W45 compliant)

The following specifications indicate the limits of particulates that the system can tolerate:

• Reactive gases:
  • The copper reactivity level shall be less than 200 Angstroms per month (Å/month)
  • The silver reactivity level shall be less than 200 Å/month
• Airborne particulates:
  • The room air should be continuously filtered with MERV 8 filters.
  • Air entering a data center should be filtered with MERV 11 or preferably MERV 13 filters.
  • The deliquescent relative humidity of the particulate contamination should be more than 60% RH
  • Environment must be free of zinc whiskers

For additional information, see the Specifications section of the documentation for the server, available from the Lenovo Documents site, https://pubs.lenovo.com/

The NVIDIA GB300 NVL72 conforms to the following standards:

• ANSI/UL 62368-1
• IEC 62368-1 (CB Certificate and CB Test Report)
• CSA C22.2 No. 62368-1
• Mexico NOM-019
• CE, UKCA Mark (EN55032 Class A, EN62368-1, EN55035, EN61000-3-11, EN61000-3-12, (EU) 2019/424, and EN IEC 63000 (RoHS))
• FCC - Verified to comply with Part 15 of the FCC Rules, Class A
• Canada ICES-003, issue 7, Class A
• CISPR 32, Class A, CISPR 35
• Japan VCCI, Class A
• Taiwan BSMI CNS15936, Class A; CNS15598-1; Section 5 of CNS15663
• Australia/New Zealand AS/NZS CISPR 32, Class A; AS/NZS 62368.1
• SGS, VOC Emission
• Japanese Energy-Saving Act
• EU2019/424 Energy Related Product (ErP Lot9)

The server and enclosure have the following warranty:

• Lenovo GB300 Compute tray (7DLZ) - 3-year warranty
• Lenovo GB300 NVL Switch Tray (7DJY) - 3-year warranty
• SN2201 Mgmt Switch (7DJW) - 3-year warranty
• Power Shelf, 1U, 33kW (C5RR) - 3-year warranty

For more information, consult the brochure Lenovo Operational Support Services for Data Centers Services.

* Some service levels may not be available in all markets. Contact your sales representative for more information.

Lenovo Data Center Services empower you at every stage of your IT lifecycle. From expert advisory and strategic planning to seamless deployment and ongoing support, we ensure your infrastructure is built for success. Our comprehensive services accelerate time to value, minimize downtime, and free your IT staff to focus on driving innovation and business growth.

In this section:

• Lenovo Advisory Services
• Lenovo Plan & Design Services
• Lenovo Deployment, Migration, and Configuration Services
• Lenovo Support Services
• Lenovo Managed Services
• Lenovo Sustainability Services
• Additional Lenovo Essential Equipment Service

Lenovo Deployment, Migration, and Configuration Services

Optimize your IT operations by shifting labor-intensive functions to Lenovo's skilled technicians for seamless on-site or remote deployment, configuration, and migration. Enjoy peace of mind, faster time to value, and comprehensive knowledge sharing with your IT staff, backed by our best-practice methodology.

• Deployment Services for Storage and ThinkAgile

  A comprehensive range of remote and onsite options tailored specifically for your business needs to ensure your storage and ThinkAgile hardware are fully operational from the start.

• Hardware Installation Services

  A full-range, comprehensive setup for your hardware, including unpacking, inspecting, and positioning components to ensure your equipment is operational and error-free for the most seamless and efficient installation experience, so you can quickly benefit from your investments.

• DM/DG File Migration Services

  Take the burden of file migration from your IT’s shoulders. Our experts will align your requirements and business objectives to the migration plans while coordinating with your team to plan and safely execute the data migration to your storage platforms.

• DM/DG/DE Health Check Services

  Our experts perform proactive checks of your Firmware and system health to ensure your machines are operating at peak and optimal efficiency to maximize up-time, avoid system failures, ensure the security of IT solutions and simplify maintenance.

• Factory Integrated Services

  A suite of value-added offerings provided during the manufacturing phase of a server or storage system that reduces time to value. These services aim at improving your hardware deployment experience and enhance the quality of a standard configuration before it arrives at your facility.

Lenovo Support Services

In addition to response time options for hardware parts, repairs, and labor, Lenovo offers a wide array of additional support services to ensure your business is positioned for success and longevity. Our goal is to reduce your capital outlays, mitigate your IT risks, and accelerate your time to productivity.

• Premier Support for Data Centers

  Your direct line to the solution that promises the best, most comprehensive level of support to help you fully unlock the potential of your data center.

• Premier Enhanced Storage Support (PESS)

  Gain all the benefits of Premier Support for Data Centers, adding dedicated storage specialists and resources to elevate your storage support experience to the next level.

• Committed Service Repair (CSR)

  Our commitment to ensuring the fastest, most seamless resolution times for mission-critical systems that require immediate attention to ensure minimal downtime and risk for your business. This service is only available for machines under the Premier 4-Hour Response SLA.

• Multivendor Support Services (MVS)

  Your single point of accountability for resolution support across vast range of leading Server, Storage, and Networking OEMs, allowing you to manage all your supported infrastructure devices seamlessly from a single source.

• Keep Your Drive (KYD)

  Protect sensitive data and maintain compliance with corporate retention and disposal policies to ensure your data is always under your control, regardless of the number of drives that are installed in your Lenovo server.

• Technical Account Manager (TAM)

  Your single point of contact to expedite service requests, provide status updates, and furnish reports to track incidents over time, ensuring smooth operations and optimized performance as your business grows.

• Enterprise Software Support (ESS)

  Gain comprehensive, single-source, and global support for a wide range of server operating systems and Microsoft server applications.

For more information, consult the brochure Lenovo Operational Support Services for Data Centers.

Lenovo Managed Services

Achieve peak efficiency, high security, and minimal disruption with Lenovo's always-on Managed Services. Our real-time monitoring, 24x7 incident response, and problem resolution ensure your infrastructure operates seamlessly. With quarterly health checks for ongoing optimization and innovation, Lenovo's remote active monitoring boosts end-user experience and productivity by keeping your data center's hardware performing at its best.

Lenovo Managed Services provides continuous 24x7 remote monitoring (plus 24x7 call center availability) and proactive management of your data center using state-of-the-art tools, systems, and practices by a team of highly skilled and experienced Lenovo services professionals.

Quarterly reviews check error logs, verify firmware & OS device driver levels, and software as needed. We’ll also maintain records of latest patches, critical updates, and firmware levels, to ensure you systems are providing business value through optimized performance.

Lenovo Sustainability Services

• Asset Recovery Services

  Lenovo Asset Recovery Services (ARS) provides a secure, seamless solution for managing end-of-life IT assets, ensuring data is safely sanitized while contributing to a more circular IT lifecycle. By maximizing the reuse or responsible recycling of devices, ARS helps businesses meet sustainability goals while recovering potential value from their retired equipment. For more information, see the Asset Recovery Services offering page.

• CO2 Offset Services

  Lenovo’s CO2 Offset Services offer a simple and transparent way for businesses to take tangible action on their IT footprint. By integrating CO2 offsets directly into device purchases, customers can easily support verified climate projects and track their contributions, making meaningful progress toward their sustainability goals without added complexity.

• Lenovo Certified Refurbished

  Lenovo Certified Refurbished offers a cost-effective way to support IT circularity without compromising on quality and performance. Each device undergoes rigorous testing and certification, ensuring reliable performance and extending its lifecycle. With Lenovo’s trusted certification, you gain peace of mind while making a more sustainable IT choice.

• Data Center Power and Cooling Services

  The Data Center Infrastructure team will provide solution design and implementation services to support the power and cooling needs of the multi-node chassis and multi-rack solutions. This includes designing for various levels of power redundancy and integration into the customer power infrastructure. The Infrastructure team will work with site engineers to design an effective cooling strategy based on facility constraints or customer goals and optimize a cooling solution to ensure high efficiency and availability. The Infrastructure team will provide the detailed solution design and complete integration of the cooling solution into the customer data center. In addition, the Infrastructure team will provide rack and chassis level commissioning and stand-up of the water-cooled solution which includes setting and tuning of the flow rates based on water temperature and heat recovery targets. Lastly, the Infrastructure team will provide cooling solution optimization and performance validation to ensure the highest overall operational efficiency of the solution.

Additional Lenovo Essential Equipment Service

For safety reasons, it is highly recommended to use the Genie® Lift™ GL™-8 due to the server's weight. Lenovo offers the lift with add on fixtures, see Genie Material Lift for more information. If no lift is available onsite, customers must move the machine to an accessible, powered area before the technician arrives and handle reinstallation.

Lenovo strongly advises configuring a complete solution for GB300 and SC-Systems with the lift tool through options.

Essential Equipment that must be available at the customer site ahead of each Service event.

• Service Lift: PN: 4XF7B02087
• Air compressor:
  • Max Pressure: 150 PSI
  • Air Flow: 1.41 SCFM at 0 PSI
  • Accuracy: +/- 1 PSI with auto-pressure check
  • Hose: 36-inch hose with all-brass chuck
  • Smart Features: Auto-shutoff at target PSI and 4 programmable memory presets
  • System: Cordless - Battery operated - with charger and battery
• Fill charger:
  • Flow Rate: 480 Gallons Per Hour (GPH)
  • Connections: 3/4" Brass (Standard garden hose threads)
  • Lift/Head: 18 ft. max suction lift; 75 ft. max head height
  • Priming: Self-priming (No manual priming required)
  • Protection: Auto-shutoff when water stops flowing (prevents motor damage)
  • System: Cordless - Battery operated - with charger and battery
• 2x 30L Labeled tanks for new and discarded liquid coolant
• Spill kit
• PG25 or NALCO® TCS101 or HPCCL2000 (DI water premixed with antimicrobial agent)
• Service kit (FRU PN: 03PV563/L2 PN: STL7C32889)

Notes:

Onsite spares (compute nodes) would be needed to support an onsite Next Business Day Service Level target. 2% of the order recommended quantity. For Compute Tray base alternatives available for selection, see the Spare Compute Trays section.

The  GB300 Compute trays, GB300 NVL Switch Trays, SN2201 Management Switches and Power Shelves are supported in the following racks:

• 7DJVCTO2WW Lenovo 48U MGX Rack

Considering the weight of the trays in the server, an onsite material lift is required to allow service by a single person. If you do not already have a material lift available, Lenovo offers the Genie Lift GL-8 material lift as configurable option to the rack cabinets. Ordering information is listed in the following table.

Note: If neither the Genie Lift GL-8 nor the ServerLift SL-350x is available onsite when onsite service is required, the customer will be responsible for getting the system to a suitable work surface (with access to power) prior to service technician arrival and returning the system to the rack when service is complete prior to service technician departure.

Why wait to obtain the technology you need now? No payments for 90 days and predictable, low monthly payments make it easy to budget for your Lenovo solution.

• Flexible

  Our in-depth knowledge of the products, services and various market segments allows us to offer greater flexibility in structures, documentation and end of lease options.

• 100% Solution Financing

  Financing your entire solution including hardware, software, and services, ensures more predictability in your project planning with fixed, manageable payments and low monthly payments.

• Device as a Service (DaaS)

  Leverage latest technology to advance your business. Customized solutions aligned to your needs. Flexibility to add equipment to support growth. Protect your technology with Lenovo's Premier Support service.

• 24/7 Asset management

  Manage your financed solutions with electronic access to your lease documents, payment histories, invoices and asset information.

• Fair Market Value (FMV) and $1 Purchase Option Leases

  Maximize your purchasing power with our lowest cost option. An FMV lease offers lower monthly payments than loans or lease-to-own financing. Think of an FMV lease as a rental. You have the flexibility at the end of the lease term to return the equipment, continue leasing it, or purchase it for the fair market value. In a $1 Out Purchase Option lease, you own the equipment. It is a good option when you are confident you will use the equipment for an extended period beyond the finance term. Both lease types have merits depending on your needs. We can help you determine which option will best meet your technological and budgetary goals.

Ask your Lenovo Financial Services representative about this promotion and how to submit a credit application. For the majority of credit applicants, we have enough information to deliver an instant decision and send a notification within minutes.

The following sales training courses are offered for employees and partners (login required). Courses are listed in date order.

1. FY26Q4 Solutions Launch GTC Launch Quick Hit
   2026-03-12 | 10 minutes | Employees and Partners
   Details
   FY26Q4 Solutions Launch GTC Launch Quick Hit
   
   

   This Quick Hit covers the Lenovo Hybrid AI Advantage - from personal AI to gigawatt scale AI factories. Lenovo is uniquely positioned to meet your customers' needs in all AI environments, no matter how small or large. The combination of hardware, software, services, and solutions is unmatched in the industry.

   Published: 2026-03-12
   Length: 10 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: DAIQ101
2. HPC VTT: Unlocking Hybrid HPC + AI_ Slinky Bridge for Unified GPU Workloads
   2026-03-10 | 62 minutes | Employees Only
   Details
   HPC VTT: Unlocking Hybrid HPC + AI_ Slinky Bridge for Unified GPU Workloads
   
   

   View this session to hear from our speakers, Aurelien Ortiz, Software Architect
   HPC & AI, Lenovo and Nick Ihli, Sr Product Manager System Software – Slurm, SchedMD, as they explain how Slinky helps with unified GPU workloads. Topics include:
   Why Slinky? Bridging HPC and Cloud-Native AI Workloads
   How Slinky Bridge Works: Architecture, Components, and Flow
   Deployment Requirements & How to Position Slinky

   Published: 2026-03-10
   Length: 62 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   

   Course code: DVHPC231
3. Edge VTT - NVIDIA Robotics Platform
   2026-01-08 | 67 minutes | Employees Only
   Details
   Edge VTT - NVIDIA Robotics Platform
   
   

   In this session we feature speakers from both NVIDIA and Lenovo. Attendees will learn about NVIDIA’s platform stack for Robotics and what Lenovo is doing in the field of robotics.
   
   During this session we will dive into NVIDIA’s three-computer stack for Physical AI. Our speaker will explore libraries and workflows to develop, train, simulate, deploy, operate, and optimize AI robot systems and software. This session will cover the basics of the technical platform, how to get started and case studies from some NVIDIA's ecosystem.
   
   Objectives:
   Discuss acceleration libraries
   Describe simulation workflows
   List foundational models for robotics
   
   Tags: Artificial Intelligence (AI), Sales, Software Platforms, Technical Sales

   Published: 2026-01-08
   Length: 67 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   

   Course code: DVEDG223
4. Lenovo Unlocks the Power of NIM: Overview, Industry Use Cases and Lenovo Services
   2025-10-16 | 55 minutes | Employees and Partners
   Details
   Lenovo Unlocks the Power of NIM: Overview, Industry Use Cases and Lenovo Services
   
   

   Join us for an insightful session with our Lenovo speakers, Farah Toos and Dinesh Tripathi where we’ll explore the transformative potential of NVIDIA Inference Microservices (NIM). This webinar will provide a comprehensive overview of NIMs, highlighting how it streamlines operations, enhances scalability, and drives innovation across industries.
   
   Discover real-world use cases in sectors such as healthcare, manufacturing, retail, and finance, and learn how Lenovo’s portfolio of services—including deployment, optimization, and lifecycle support—can help your customers maximize the value of their infrastructure investments.
   Whether you're engaging with enterprise clients or mid-market opportunities, this session will equip you with the knowledge and tools to position Lenovo’s NIM solutions effectively and drive impactful conversations.
   
   Key Takeaways:
   - Understand the core capabilities and benefits of NIM
   - Explore industry-specific applications and success stories
   - Learn how Lenovo services complement and enhance NIM deployments
   - Gain selling strategies and resources to support customer engagements
   
   Tags: Artificial Intelligence (AI), NVIDIA,Services,Technical Sales, ThinkAgile, ThinkSystem

   Published: 2025-10-16
   Length: 55 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: DVCLD228
5. Partner Technical Webinar - NVIDIA Software
   2025-07-21 | 60 minutes | Employees and Partners
   Details
   Partner Technical Webinar - NVIDIA Software
   
   

   In this 60-minute replay, Carlos Huescas, Lenovo, and Sandeep Brahmarouthu and Rob Magno of NVIDIA, presented the key software offerings of NVDIA AI Enterprise (NVAIE) and Run:ai, including a demo of Run:ai.
   
   Tags: Artificial Intelligence (AI)

   Published: 2025-07-21
   Length: 60 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: JUL1825
6. Partner Technical Webinar - AI Vertical Spotlight Pt 2
   2025-07-08 | 60 minutes | Employees and Partners
   Details
   Partner Technical Webinar - AI Vertical Spotlight Pt 2
   
   

   In this 60-minute replay, we concluded the AI Vertical Spotlight (Pt 2) with our final two speakers. Peter Orban, AI Business Development Manager, discussed Financial and Banking, while Eric Skomra, Public Sector & Spaces AI Technologist, provided insights on State, Local, Education (SLED), and Smart Spaces.
   
   Tags: Artificial Intelligence (AI)

   Published: 2025-07-08
   Length: 60 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: JUN2725
7. AI VTT: NVIDIA Run:ai
   2025-07-02 | 75 minutes | Employees Only
   Details
   AI VTT: NVIDIA Run:ai
   
   

   NVIDIA Run:ai is a GPU orchestration and optimization platform designed to help organizations maximize their GPU compute resources for AI workloads. It accelerates AI development, reduces costs, and improves AI development cycles by enabling dynamic allocation and scheduling of GPU resources, as well as workload submission and sharing. Essentially, it provides a centralized interface to manage AI compute infrastructure, making it easier for AI teams to access and utilize GPUs effectively.
   
   Join Carlos Huescas from Lenovo, Sandeep Brahmarouthu and Robert Magno from NVIDIA as they discuss NVIDIA Run:ai. Topics include:
   •What is Run:ai and its capabilities?
   •Customer segmentation for Run:ai
   •How to order, part numbers and licensing
   •Demo of Run:ai
   
   Tags: Artificial Intelligence (AI), NVIDIA

   Published: 2025-07-02
   Length: 75 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   

   Course code: DVAI218
8. Partner Technical Webinar - Enterprise AI Team Intro and Vertical Spotlight Pt1
   2025-06-17 | 60 minutes | Employees and Partners
   Details
   Partner Technical Webinar - Enterprise AI Team Intro and Vertical Spotlight Pt1
   
   

   In this 60-minute replay, John Encizo introduced his new Enterprise AI Team. Part 1 covered three verticals: Retail with Allen Holmes, Manufacturing with Jason Hamp, and Healthcare with Janna Templin.
   
   Tags: Artificial Intelligence (AI)

   Published: 2025-06-17
   Length: 60 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: JUN1325
9. VTT Edge: Understanding Visual AI Agents with NVIDIA June 2025
   2025-06-16 | 60 minutes | Employees and Partners
   Details
   VTT Edge: Understanding Visual AI Agents with NVIDIA June 2025
   
   

   Join our guest speakers from NVIDIA as they discuss what’s behind the scenes of visual AI Agents for Smart Cities, Smart Spaces and Manufacturing. Explore the modular approach to building a workforce of AI Agents. Topics include:
   
   • Sensors which feed the AI Agents
   • How AI agents improve safety and prevent accidents in Smart Spaces
   • Demo: Modular development of AI Agents
   
   Tags: Artificial Intelligence (AI), Technical Sales, NVIDIA

   Published: 2025-06-16
   Length: 60 minutes
   

   Start the training:
   Employee link: Grow@Lenovo
   Partner link: Lenovo 360 Learning Center
   

   Course code: DVEDG221
10. Lenovo Cloud Architecture VTT: Supercharge Your Enterprise AI with NVIDIA AI Enterprise on Lenovo Hybrid AI Platform
    2025-04-17 | 75 minutes | Employees and Partners
    Details
    Lenovo Cloud Architecture VTT: Supercharge Your Enterprise AI with NVIDIA AI Enterprise on Lenovo Hybrid AI Platform
    
    

    Join us for an in-depth webinar with Justin King, Principal Product Marketing Manager for Enterprise AI exploring the power of NVIDIA AI Enterprise, delivering Generative and Agentic AI outcomes deployed with Lenovo Hybrid AI platform environments.
    In today’s data-driven landscape, AI is evolving at high speed, with new techniques delivering more accurate responses. Enterprises are seeking not just an understanding but also how they can achieve AI-driven business outcomes.
    With this, the demand for secure, scalable, and high-performing AI operations-and the skills to deliver them-is top of mind for many. Learn how NVIDIA AI Enterprise, a comprehensive software suite optimized for NVIDIA GPUs, provides the tools and frameworks, including NVIDIA NIM, NeMo, and Blueprints, to accelerate AI development and deployment while reducing risk-all within the control and security of your Lenovo customer’s hybrid AI environment.
    
    Tags: Artificial Intelligence (AI), Cloud, Data Management, Nvidia, Technical Sales

    Published: 2025-04-17
    Length: 75 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    Partner link: Lenovo 360 Learning Center
    

    Course code: DVCLD221
11. AI VTT: GTC Update and The Lenovo LLM Sizing Guide
    2025-03-12 | 86 minutes | Employees Only
    Details
    AI VTT: GTC Update and The Lenovo LLM Sizing Guide
    
    

    Please view this session that is two parts. Part one is Robert Daigle, Director, Global AI Solutions and Hande Sahin-Bahceci, AI Solutions Marketing Leader explaining the upcoming announcements for NVIDIA GTC. Part Two is Sachin Wani, AI Data Scientist explaining the Lenovo LLM Sizing Guide with these topics:
    
    • Minimum GPU requirements for fine-tuning/training and inference
    • Gathering requirements for the customer's use case
    • LLMs from a technical perspective
    
    Tags: Artificial Intelligence (AI), Technical Sales

    Published: 2025-03-12
    Length: 86 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    

    Course code: DVAI214
12. VTT AI: Components of the AI Stack and Where Lenovo Sits November 2024
    2024-11-26 | 75 minutes | Employees Only
    Details
    VTT AI: Components of the AI Stack and Where Lenovo Sits November 2024
    
    

    Join Per Ljungstrom, Lenovo Principal TC EMEA, as he explores AI concepts where innovations meet simplified predefined solutions which deploy at scale. Topics for this session include:
    • Associating software with the ground level of hardware
    • Attach NVIDIA AI Enterprise, Microsoft, Tiber AI Stacks and more
    • AI at the Edge and the complete solution
    • What to consider when talking AI Stack with your customer
    
    Tags: Artificial Intelligence (AI), Cloud, Technical Sales, Technology solutions, ThinkEdge

    Published: 2024-11-26
    Length: 75 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    

    Course code: DVAI210
13. VTT AI: NVIDIA OVX
    2024-10-23 | 55 minutes | Employees and Partners
    Details
    VTT AI: NVIDIA OVX
    
    

    Please join this session as Steven Puzio, Global Sales Leader of NVIDIA Omniverse speaks to us about these topics:
    
    • OVX use cases
    • Target customers
    • OVX reference architectures
    • Parts, pieces and technical details
    
    Tags: Artificial Intelligence (AI), Nvidia

    Published: 2024-10-23
    Length: 55 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    Partner link: Lenovo 360 Learning Center
    

    Course code: DVAI209
14. Think AI Weekly: Ride the NVIDIA Wave for AI
    2024-10-07 | 60 minutes | Employees Only
    Details
    Think AI Weekly: Ride the NVIDIA Wave for AI
    
    

    In this session, a panel including speakers from NVIDIA, Lenovo IDG and Lenovo ISG address the topics:
    •Leveraging AI workstations to start an AI journey
    •Leading an ISG sale with NVIDIA AI Enterprise
    •NVIDIA sales tools available for Lenovo sellers
    •NVIDIA training on grow@lenovo and more
    
    Tags: Artificial Intelligence (AI), Nvidia

    Published: 2024-10-07
    Length: 60 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    

    Course code: DTAIW121
15. Lenovo VTT Cloud Architecture - Unlock Gen AI with VMware Private AI Foundation with NVIDIA
    2024-07-16 | 60 minutes | Employees and Partners
    Details
    Lenovo VTT Cloud Architecture - Unlock Gen AI with VMware Private AI Foundation with NVIDIA
    
    

    In today's rapidly evolving digital landscape, businesses are hungry for the transformative power of Artificial Intelligence (AI). They see AI as the key to streamlining operations and unlocking exciting new opportunities. However, widespread adoption has been hampered by concerns surrounding privacy, the complexity of implementation, and the hefty costs associated with deploying and managing AI solutions at an enterprise level.
    Join Chris Gully and Baker Hull, Solutions Architects from VMware by Broadcom, as they discuss how Lenovo, NVIDIA, and VMware By Broadcom are partnering to deliver a private, secure, scalable, and flexible AI infrastructure solution that helps enterprise customers build and deploy AI workloads within their own private cloud infrastructure, ensure the control of sensitive data and compliance with regulatory requirements, ultimately driving faster time to value and achieving their AI objectives.
    
    Tags: Artificial Intelligence (AI), Cloud, Nvidia, ThinkAgile, VMware

    Published: 2024-07-16
    Length: 60 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    Partner link: Lenovo 360 Learning Center
    

    Course code: DVCLD214
16. Guidance for Selling NVIDIA Products at Lenovo for ISG
    2024-07-01 | 25 minutes | Employees and Partners
    Details
    Guidance for Selling NVIDIA Products at Lenovo for ISG
    
    

    This course gives key talking points about the Lenovo and NVIDIA partnership in the Data Center. Details are included on where to find the products that are included in the partnership and what to do if NVIDIA products are needed that are not included in the partnership. Contact information is included if help is needed in choosing which product is best for your customer. At the end of this session sellers should be able to explain the Lenovo and NVIDIA partnership, describe the products Lenovo can sell through the partnership with NVIDIA, help a customer purchase other NVIDIA product, and get assistance with choosing NVIDIA products to fit customer needs.
    
    Tags: Artificial Intelligence (AI), Nvidia

    Published: 2024-07-01
    Length: 25 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    Partner link: Lenovo 360 Learning Center
    

    Course code: DNVIS102
17. NVIDIA AI Solutions and Market Trends
    2023-10-12 | 55 minutes | Employees Only
    Details
    NVIDIA AI Solutions and Market Trends
    
    

    The purpose of this course is to help the learner recognized AI Market and trends. Also, explain NVIDIA's Computing platform, and discuss its importance for the market.
    
    Course Objectives:
    Recognize AI Trends
    Explain NVIDIA Computing Platform
    Discuss Industry Verticals Marketing
    
    Tags: Artificial Intelligence (AI), Nvidia, Sales

    Published: 2023-10-12
    Length: 55 minutes
    

    Start the training:
    Employee link: Grow@Lenovo
    

    Course code: DAINVD101

1-10 of 17 courses. Show All

For more information, see these resources:

• Lenovo NVIDIA GB300 NVL72 server product page
  https://www.lenovo.com/us/en/servers-storage/?IPromoID=LEN781264
• Lenovo NVIDIA GB300 NVL72 datasheet
• https://lenovopress.lenovo.com/datasheet/DS0207
• Interactive 3D Tour of the Lenovo NVIDIA GB300 NVL72:
• https://lenovopress.lenovo.com/datasheet/LP2381
• Lenovo NVIDIA GB300 NVL72 drivers and support
  https://datacentersupport.lenovo.com/us/en/products/solutions-and-software/aiinfrastructure/gb300nvl72/7djv/downloads/driver-list/

• ServerProven hardware compatibility:
  https://serverproven.lenovo.com/
• Data Center Solution Configurator (DCSC)
  https://dcsc.lenovo.com
• Lenovo Cluster solutions configurator (x-config)
  https://lesc.lenovo.com/products/hardware/configurator/worldwide/bhui/asit/index.html