Lenovo ThinkStation PGX AI Workstation

NVIDIA Grace Blackwell GB10 | Petaflop-Class AI Performance

The Lenovo ThinkStation PGX represents a new class of AI-first professional workstations, purpose-built to deliver petaflop-class AI compute performance anywhere. Powered by the groundbreaking NVIDIA® Grace Blackwell GB10 Superchip, this workstation is engineered for AI developers, data scientists, research institutions, enterprise innovation teams, and advanced edge-AI deployments that demand extreme computational density, ultra-high memory bandwidth, and enterprise-grade networking in a compact, power-efficient form factor.

Built around a 20-core ARM-based NVIDIA Grace CPU tightly integrated with Blackwell GPU technology, the ThinkStation PGX delivers unprecedented performance for AI model development, inference, simulation, data analytics, and accelerated computing workloads. With 128GB of ultra-fast LPDDR5X memory, 400 Gigabit Ethernet via NVIDIA ConnectX-7 Smart NIC, and NVIDIA DGX OS, this platform brings datacentre-class AI capability directly to the desktop or edge environment.

Available through 7Stores.au, the Lenovo ThinkStation PGX is supplied as genuine Australian stock, backed by Premier Next Business Day and On-site warranty, making it an ideal solution for organisations investing in next-generation AI infrastructure without the complexity of full datacentre deployments.

NVIDIA Grace Blackwell GB10 – AI Compute Without Compromise

At the core of the ThinkStation PGX is the NVIDIA® GB10 Grace Blackwell Superchip, delivering petaflop-level AI performance within a compact workstation footprint. This architecture is designed to remove traditional bottlenecks between CPU, GPU, and memory, enabling massive parallelism, ultra-low latency data access, and extreme AI throughput.

The 20-core ARM-based Grace processor provides exceptional efficiency and scalability for AI pipelines, large language models, training, inference, and simulation workloads, allowing teams to iterate faster, deploy smarter models, and achieve results previously limited to large DGX systems in datacentres.
This makes the ThinkStation PGX an ideal complement to enterprise AI datacentre environments, enabling local development, testing, and edge deployment while maintaining architectural consistency with NVIDIA DGX platforms.

128GB LPDDR5X – Massive Unified Memory for AI Workloads

The ThinkStation PGX is equipped with 128GB of LPDDR5X memory, delivering exceptionally high bandwidth and low latency, which is critical for AI model execution, large dataset manipulation, and high-concurrency workloads. This large unified memory pool enables seamless multitasking and ensures that complex AI models and datasets remain in memory, significantly reducing data transfer overhead and improving overall system efficiency.
The use of LPDDR5X technology ensures faster data transfer rates, lower power consumption, and improved performance per watt—perfectly aligned with the system’s 170W power envelope.

Enterprise-Grade Networking with 400 Gigabit Ethernet

Designed for modern AI environments, the ThinkStation PGX integrates NVIDIA ConnectX-7 Smart NIC, delivering 400 Gigabit Ethernet connectivity. This enables ultra-high-speed data transfer, low-latency communication, and seamless integration into AI clusters, storage networks, and high-performance computing environments.
Whether deployed as a standalone AI workstation or as part of a distributed AI infrastructure, the PGX ensures reliable, high-throughput connectivity for data-intensive workloads.

NVIDIA DGX OS – Datacentre Intelligence on the Desktop

The ThinkStation PGX runs NVIDIA DGX OS, a Linux-based operating system optimised for AI, machine learning, and accelerated computing. DGX OS is engineered to efficiently manage hardware resources, optimise GPU utilisation, and provide a stable platform for AI frameworks and development tools.
By aligning with NVIDIA’s DGX software stack, the PGX enables consistent development, testing, and deployment workflows between desktop, edge, and datacentre environments—dramatically reducing friction and accelerating AI innovation.

Compact Design with Exceptional Efficiency

Despite its extreme AI capability, the ThinkStation PGX operates within a 170W power consumption envelope, making it one of the most power-efficient petaflop-class AI systems available. Its compact black chassis is designed for office, lab, studio, and edge deployments, eliminating the need for specialised datacentre infrastructure while maintaining enterprise reliability.

Why Buy the Lenovo ThinkStation PGX from 7Stores.au