Lenovo is pushing a compact desktop into serious AI territory with the ThinkStation PGX, now available in Indonesia. The workstation is positioned for professionals who need high-end AI computing without moving to a large server setup.
At the center of the device is NVIDIA GB10 Grace Blackwell Superchip, which gives the ThinkStation PGX up to 1 PetaFlop, or 1000 TOPS, of AI performance. Lenovo says the system is designed to handle modern generative AI workloads and can run models with up to 200 billion parameters.
Built for professional AI work
The ThinkStation PGX targets AI developers, researchers, data scientists, students, and engineers. Lenovo is presenting it as a personal AI workstation that fits into a compact desktop form while still supporting demanding tasks such as prototyping, fine-tuning, and inference for reasoning models.
To keep those workloads moving efficiently, the system comes with 128 GB of unified system memory. Lenovo also specifies memory bandwidth of 273 GB/s, which is meant to help maintain steady data flow when working with large datasets.
Scalability beyond a single unit
For users who need more capacity, Lenovo offers a way to scale the setup further. Two ThinkStation PGX units can be connected to support AI models of up to 405 billion parameters.
That approach is aimed at professional environments where workloads can grow quickly. It also gives organizations an alternative to relying more heavily on cloud infrastructure or large-scale servers.
Software stack ready from the start
Another key part of the ThinkStation PGX is its software environment. The workstation ships with NVIDIA DGX OS and the NVIDIA AI software stack, including widely used tools such as PyTorch and Jupyter.
Lenovo says this setup allows users to develop, test, and run AI models on the desktop before moving them to a data center or cloud environment. The company also says the workflow can be done without additional code changes, which should make deployment faster and more practical for teams.
Connectivity for heavy data transfer
The hardware also includes features meant to support AI work that depends on fast data movement. ThinkStation PGX offers four USB4 ports with speeds of up to 20Gb/s, along with 10 GbE LAN connectivity.
Lenovo also adds support for NVIDIA ConnectX-7, which enables two PGX units to be linked together. This is part of the machine’s broader focus on scalability and high-performance AI processing.
Lenovo’s broader AI workstation push
In Indonesia, Lenovo continues to expand its AI-ready workstation lineup through the ThinkStation and ThinkPad P Series. The company says its Lenovo Hybrid AI Advantage with NVIDIA is designed to accelerate the development and deployment of agentic AI from the edge, data center, and cloud.
Susianty Bong, SMB One Channel Lead Lenovo Indonesia, said the collaboration with NVIDIA is intended to deliver high-performance AI devices in an efficient and accessible design. She also said the launch supports Lenovo’s Smarter AI for All commitment, with the goal of making generative AI more widely usable in Indonesia.
ThinkStation PGX is now available in Indonesia through the Lenovo Official Store and Authorized Store. Lenovo also offers customization options so the final price depends on the configuration selected.