Lenovo PGX Workstation, NVIDIA GB10, 128GB RAM, 4TB SSD, DGX OS

The Lenovo PGX Workstation is a high-performance, compact AI computing solution designed to bring advanced model development and testing directly to your desktop. Featuring the powerful NVIDIA GB10 Grace Blackwell superchip, this workstation provides the computational capacity necessary for data scientists, researchers, and software developers to prototype and fine-tune large language models locally. By eliminating the need for constant cloud reliance, it offers a secure, cost-effective, and energy-efficient environment for complex AI workflows.

Advantages

This workstation excels by providing massive AI compute power in a small form factor, delivering up to 1000 TOPS of performance. The integration of 128GB of unified LPDDR5x memory ensures seamless handling of large datasets and complex parameters. With the pre-installed NVIDIA DGX OS and a comprehensive software stack including CUDA 13 and AI Workbench, users can transition immediately from setup to development. Furthermore, the ability to daisy-chain two units allows for scaling up to 405-billion-parameter models, providing unparalleled flexibility for professional AI research.

Suitable for

• Local development and fine-tuning of large language models
• Data science research and complex simulation environments
• Prototyping AI applications without cloud dependency
• Secure on-premises AI model testing

When it may not be suitable

This system is specifically engineered for AI development and computational research, making it less ideal for general office tasks, standard business productivity, or high-end gaming. Users requiring a traditional Windows-based environment or extensive peripheral expansion beyond the provided USB4 and networking ports may find this specialized hardware restrictive.

Key specifications

• NVIDIA GB10 Grace Blackwell AI superchip
• 20-core Arm processor architecture
• 128GB LPDDR5x unified memory
• 4TB M.2 PCIe 4.0 NVMe SSD
• 10GbE RJ45 and WiFi 7 connectivity
• NVIDIA DGX OS based on Ubuntu Linux Pro
• Dimensions: 150x150x50.5mm
• Weight: 1.2kg

How to use

To begin, connect the workstation to a power source using the included 240W USB-C adapter and attach your display via the USB4 or HDMI ports. Upon powering on, the pre-installed NVIDIA DGX OS will guide you through the initial configuration. Utilize the pre-loaded NVIDIA AI software stack, including PyTorch and Jupyter, to start your development projects. For advanced scaling, use the QSFP ConnectX-7 ports to link two PGX units together for increased model parameter support.

What to consider before purchase

• Ensure your software requirements are compatible with the Linux-based NVIDIA DGX OS.
• Verify that your workspace supports the specific connectivity options provided, such as USB4 and 10GbE.
• Consider if your current AI projects require the local processing power of the GB10 chip versus cloud-based alternatives.
• Review the security features like TPM 2.0 and self-encrypting drives to ensure they meet your organizational compliance standards.

Usage recommendations

For optimal performance and longevity, ensure the workstation is placed in a well-ventilated area to maintain thermal efficiency during intensive AI training tasks. Regularly update the NVIDIA AI software stack and firmware via the provided management tools to benefit from the latest security patches and performance optimizations.