Acer announced that its Veriton GN100 AI Mini Workstation served as the official hardware platform for The Spark Hack Series – New York presented by NVIDIA. There are several new capabilities of Veriton GN100, including support for NVIDIA NemoClaw and the new Acer Sense Pro platform for centralized system management and added AI developer tools. The workstation also now enables connecting up to four systems for running larger AI models of up to 700 billion parameters.
Acer Veriton GN100 Supports NVIDIA NemoClaw for Secure, Autonomous AI Agent Development
Built on NVIDIA DGX Spark platform with NVIDIA GB10 Grace Blackwell Superchip, the Acer Veriton GN100 brings server-class AI performance to an ultra-compact, power-efficient desktop designed for the next generation of AI development. With the NVIDIA AI software stack, it gives developers an easy path to prototype, fine-tune, and deploy AI agents locally, then scale seamlessly to the cloud. Delivering up to 1 PetaFLOP of FP4 AI performance and 128 GB of unified memory, the Veriton GN100 is an ideal workstation for running and deploying NemoClaw, NVIDIA’s standard for secure AI agentic workflow development.
NVIDIA NemoClaw uses open source models—like NVIDIA Nemotron—alongside the NVIDIA OpenShell runtime, to simplify and secure personal AI agent development.
OpenShell enables developers to run a broad range of long-running autonomous agentic AI systems—the next wave of AI development —inside secure development sandboxes, without modifying application code, including popular coding agents and agent platforms such as Claude Code, Codex, Cursor, and OpenCode.
These agents can go beyond completing individual tasks: they can autonomously plan multi-step work, invoke specialized sub-agents, and learn to use advanced technical tools through configuration and workflow design. This opens the door for companies to equip employees with highly specialized AI capabilities across a wide range of workflows, from spreadsheets and presentation software to CAD, 3D modeling, and visual production tools.
Security and control are built in from the start. Agents can be configured to begin with zero permissions, and developers explicitly approve access based on intent and governance policies. This gives teams precise control over what agents can do, while supporting private inference environments and cost-efficient deployment models, including local or hybrid inference, to reduce reliance on per-token cloud execution.
For more information, please visit NemoClaw playbook at build.nvidia.com or NemoClaw on LIVE at GitHub.
The Veriton GN100 now supports up to four connected systems using a RoCE 200 GbE switch, doubling the previous limit of two, enabling developers to run larger AI models—up to 700 billion parameters versus 405 billion before. The result is more capacity for advanced local inference, multi-agent workflows, and complex AI development, while preserving the privacy, control, and cost advantages of on-premises deployment.
Event participants will also have the chance to experiment with Acer’s newly launched platform Acer Sense Pro, equipped on the Veriton GN100. The proprietary system management software provides a clear, centralized way to monitor and troubleshoot systems through a single streamlined interface. Combining real-time visibility, hardware tuning, and AI-powered technical support, the application enables teams to track CPU, GPU, storage, and memory through a professional dashboard for convenient insight into overall system health and performance trends.
Acer Sense Pro also provides inferencing performance benchmarks, including Token per Second (TPS) and Time to First Token (TTFT), displayed in intuitive vertical bar charts for clear, at-a-glance analysis. Its generated context benchmark tools help developers balance intelligence and perceived latency, enabling them to fine-tune scenario-based user experiences for the best fit.
In addition, built-in self-diagnostic tools, a community-driven knowledge base, and a local AI agent help identify, explain, and quickly resolve hardware issues privately, while enhanced benchmarking capabilities help accelerate AI development by making model evaluation faster and more actionable. Users can seamlessly compare multiple AI models to assess speed and output quality, making it easier to select the most effective engine for each task.
The Acer Veriton GN100 AI Mini Workstation is available now globally. Exact specifications, prices, and availability will vary by region.
