Acer has unveiled the Veriton GN100, a compact workstation designed to run large-scale AI models locally rather than relying solely on cloud infrastructure. The device will be available in the EMEA region starting at USD 3,999.
The Veriton GN100 is powered by NVIDIA’s new GB10 Grace Blackwell Superchip, which combines CUDA and Tensor Cores with 20 Arm-based CPU cores. According to Acer, the system can reach up to 1 PFLOPS of FP4 AI performance, with 128 GB of unified memory and 4 TB of NVMe SSD storage.
Unlike traditional workstations, the GN100 is built to support on-device prototyping and fine-tuning of AI models. It ships with the NVIDIA AI software stack and supports widely used frameworks such as PyTorch, Jupyter, and Ollama. The company says this allows developers, researchers, and students to work with language models locally and scale to cloud-based systems when needed.
For larger workloads, two GN100 units can be connected via an NVIDIA ConnectX-7 SmartNIC, enabling support for models with up to 405 billion parameters. Connectivity options include Wi-Fi 7, four USB-C ports, HDMI, and Ethernet, while physical security is provided through Kensington lock support.
The GN100’s compact form factor positions it as an alternative to traditional rack-mounted servers or high-end desktops for organizations and individuals who want AI capabilities in a smaller, more space-efficient package.
Exact regional pricing and availability will vary. Acer has directed customers to check www.acer.com for local information.
