Acer brings NVIDIA Grace Blackwell to the desktop: UK early-access opens for Veriton GN100 AI Mini Workstation
Acer today invited UK organisations to register their interest in the forthcoming Veriton GN100 AI Mini Workstation, a compact professional desktop designed to bring on-device AI acceleration to engineers, data teams and creators. Powered by the NVIDIA® GB10 Grace Blackwell Superchip, the GN100 is engineered to deliver low-latency performance, stronger data privacy and more predictable costs by running common AI workloads locally close to where data lives and decisions are made.
Delivering up to 1 PFLOPS of FP4 AI performance with 128 GB unified memory and up to 4 TB self-encrypting NVMe storage, Veriton GN100 brings server-class capability to a desktop-sized device (150 × 150 × 50.5 mm; ~1.2 kg). It ships with NVIDIA DGX OS and the NVIDIA AI software stack, with support for familiar tools including PyTorch, Jupyter and Ollama, enabling rapid prototyping, fine-tuning and local deployment of LLMs and multimodal workloads.
For scale-out tasks, two GN100 units can be linked via NVIDIA® ConnectX®-7 SmartNIC to handle models up to 405 billion parameters, a flexible path to cluster-like performance at the edge or in smaller labs without data-centre build-outs.
As AI moves from pilots to production, many UK teams face the same friction points: round-trip latency to the cloud, concerns about sharing sensitive data externally, and volatile usage fees as inference scales.
The Veriton GN100 addresses these challenges head-on. Its edge-first design keeps everyday AI tasks, such as code assistance, data preparation, generative and editing workflows, and computer vision, snappy and responsive on the desktop, while helping organisations retain control of confidential datasets and right-size their use of cloud resources.
“UK businesses want AI that’s fast, private and easy to deploy,” said Craig Booth, Country Manager, Acer UK. “The Veriton GN100 brings serious, energy-efficient AI acceleration to a remarkably compact form factor, so teams get real-time responsiveness without sending everything to the cloud, and IT can manage it just like the rest of the fleet.”
Benefits at a glance for UK organisations
• Real-time responsiveness: On-device inference reduces latency for everyday AI tasks, even in bandwidth-constrained sites.
• Privacy & governance by default: Keeping data local supports UK GDPR compliance and client confidentiality requirements.
• Predictable cost: Shift routine inference from pay-per-use models to a desktop footprint for clearer TCO; use cloud only where it adds value.
• Hybrid by design: Run the right workload in the right place, local when you need speed or data control; cloud when you need elastic scale.
• Office-friendly footprint: A small-form-factor workstation suited to modern offices, studios, labs and secure/branch locations.
• Enterprise-ready security & management: Built to support modern OS and hardware security features (e.g., device encryption, secure boot, TPM integration) and to slot into existing endpoint management and identity frameworks, no AI “island” required.
For UK channel partners, the GN100 opens services-led opportunities across environment sizing, RAG and vector pipeline design, model selection/fine-tuning, secure edge deployment and lifecycle optimisation, enabling partners to package consulting, enablement and managed services around a compact, business-ready AI platform.
Pricing & availability
The Veriton GN100 AI Mini Workstation is listed from €3,999 in EMEA (and $3,999 in North America). Exact specifications, prices and UK availability timings may vary by region; UK channel and enterprise customers should contact their Acer account team or authorised partners.
Register your interest (UK): https://acer.link/4oaD8fR
Key specifications (highlights)
• NVIDIA GB10 Grace Blackwell Superchip; up to 1 PFLOPS FP4
• 128 GB LPDDR5x unified memory; up to 4 TB self-encrypting NVMe
• Wi-Fi 7, Bluetooth 5.1 or later, 4× USB 3.2 Type-C, HDMI 2.1b, Ethernet
• NVIDIA ConnectX-7 SmartNIC; Kensington lock support
• 150 × 150 × 50.5 mm, <1.5 kg
• Pre-installed NVIDIA AI stack (CUDA, cuDNN, TensorRT); supports PyTorch, Jupyter and Ollama
