- Acer Veriton GN100 uses Nvidia GB10 superchip with 128GB memory and 4TB storage
- Compact design under 1.5kg still delivers 1 PFLOPS of FP4 AI performance
- Two linked systems scale to handle AI models with up to 405 billion parameters
Acer has announced the Veriton GN100 AI Mini Workstation, a compact desktop built on Nvidia’s GB10 Grace Blackwell Superchip.
It is designed to run large AI models locally, with the option to scale workloads further by linking two systems together.
The Veriton GN100, revealed at IFA 2025 in Berlin, delivers up to 1 PFLOPS of FP4 AI performance through a combination of 20 Arm-based CPU cores, CUDA cores and fifth-generation Tensor Cores.
Up to 405 billion parameters
The device comes with 128GB of unified LPDDR5x memory and 4TB of self-encrypting NVMe storage.
Despite its diminutive 150mm x 150mm footprint, and sub-1.5kg weight, the Veriton GN100 is intended to deliver server-class compute power in a mini PC form factor.
The system includes Nvidia’s AI software stack and DGX Base OS, with support for commonly used developer tools such as PyTorch, Jupyter and Ollama.
Acer says this will allow developers, researchers and students to prototype, fine-tune and test large language models directly on the system, reducing reliance on remote cloud infrastructure and helping lower operating costs.
With an Nvidia ConnectX-7 SmartNIC installed, two Veriton GN100 units can be linked to scale workloads further, allowing processing of AI models with up to 405 billion parameters.
This will make it possible to work with even larger training sets and experiments without requiring immediate access to data center hardware.
Connectivity includes four USB 3.2 Type-C ports, HDMI 2.1b, Ethernet, Wi-Fi 7 and Bluetooth 5.1.
Security features include a Kensington lock and local execution of AI models to help protect sensitive data.
Acer has set pricing for the Veriton GN100 at $3,999 in North America and €3,999 in EMEA, with availability varying by region.
Acer is just the latest PC maker to launch a Nvidia DGX Spark clone, following in the footsteps of Asus, HP and Dell, and MSI.
