Acer is betting that enterprise AI workloads will increasingly need a home outside the cloud. Its new Veriton RI110 AI Mini Workstation combines Intel’s latest Core Ultra processor platform, Intel Arc graphics, up to 96GB of memory and OCuLink expansion in a compact desktop aimed at developers, designers and organizations running AI workloads locally.
The AI PC market is moving beyond laptops with built-in neural processing units. Acer’s latest move is toward something closer to a small-form-factor AI workstation: a desktop designed to run demanding AI applications locally while retaining the option to connect to cloud resources or external compute.
The company has introduced the Acer Veriton RI110 AI Mini Workstation (VRI110), powered by an Intel Core Ultra X7 processor 358H and Intel Arc B390 graphics. The system runs Windows 11 Pro and is designed for AI-assisted content creation, inference, software development, design and general enterprise productivity.
Acer says the machine can support AI models with up to 120 billion parameters, although real-world performance will depend heavily on model quantization, memory requirements, software optimization and workload characteristics. The claim is therefore better understood as an indication of the system’s intended model capacity rather than a guarantee of desktop-class performance across every 120-billion-parameter model.
The hardware configuration includes up to 96GB of LPDDR5X memory and a 4TB PCIe Gen 4 SSD. That combination is significant for local AI because memory capacity can become a more immediate constraint than raw processor performance when organizations attempt to run larger language models or multimodal workloads without sending sensitive data to external services.
The shift toward local AI is happening as enterprises become more selective about where workloads run. Cloud platforms from Microsoft Azure, Amazon Web Services and Google Cloud offer virtually unlimited scalability, but local inference can provide advantages around latency, data control, predictable workloads and offline operation.
Acer is positioning the Veriton RI110 as a middle ground.
Rather than replacing cloud infrastructure, the workstation can serve as an edge node for workloads that make sense to run locally while larger or more compute-intensive jobs remain in the cloud. That hybrid model is increasingly relevant as companies deploy AI agents, private models and AI-assisted development tools across business environments.
The market is already moving in this direction. Gartner forecasts that AI PCs will account for 54.7% of worldwide PC shipments in 2026, up from 31% in 2025. Gartner also expects AI PCs to become the norm by 2029.
That trend gives vendors such as Acer an incentive to differentiate AI PCs based on more than an NPU specification. Enterprise buyers increasingly need machines that can actually accommodate local AI workflows, not simply expose AI features inside conventional productivity applications.
The Veriton RI110 also comes with Copilot+ PC capabilities on Windows 11 Pro, putting it within Microsoft’s broader ecosystem of locally accelerated AI experiences. Microsoft’s approach competes with platforms from Apple and Qualcomm, while Intel continues to defend the x86 Windows enterprise market with its Core Ultra architecture and integrated AI acceleration.
Acer’s more unusual decision is to build expansion into the product through OCuLink.
The dedicated PCIe 4.0 x4 interface can provide up to 64Gbps of bandwidth for external GPUs and high-speed NVMe storage. For AI developers, this creates a potential upgrade path that is difficult to achieve with a conventional compact business PC.
An external GPU can provide substantially more acceleration for workloads that exceed the capabilities of integrated graphics, while external NVMe storage can accommodate larger model libraries and datasets.
That flexibility could matter to smaller AI teams that cannot justify purchasing a full workstation or dedicated server. A developer could begin with the integrated hardware and add external acceleration as requirements grow.
Acer is also putting an emphasis on agentic AI through its Qubi Claw software. The company’s agent runs workflows inside an isolated sandbox and is designed to allow users to deploy AI agents locally. Initial capabilities include online research, travel planning and converting audio into text notes.
The more interesting enterprise implication is the local execution model.
AI agents can potentially interact with files, applications, websites and business systems, creating a larger security surface than conventional chatbot usage. Running an agent within a sandbox does not eliminate those risks, but isolation can provide an additional control layer for organizations experimenting with autonomous workflows.
The approach also reflects a broader transition in AI computing. Gartner forecasts worldwide AI spending will reach $2.59 trillion in 2026, with AI infrastructure accounting for more than 45% of spending. The research firm expects enterprise adoption of GenAI models and AI agents to expand as organizations integrate them into broader workflows.
Acer’s mini workstation therefore arrives at an important point in the AI hardware cycle. The industry is no longer asking whether PCs should have AI capabilities. The question is increasingly where inference should happen and what level of compute enterprises need at the endpoint.
For organizations evaluating the Veriton RI110, the answer will depend on workload rather than marketing specifications. Developers running smaller quantized language models, private inference, AI-assisted coding or computer-vision applications may benefit from local compute. Teams training large foundation models will still require substantially larger GPU infrastructure.
The competitive landscape is equally broad. Dell, HP and Lenovo offer enterprise workstations and AI PCs, while NVIDIA-based systems remain the standard for many demanding AI development workloads. Apple’s high-memory Mac systems offer another alternative for developers prioritizing local model execution.
Acer’s proposition is different: a compact enterprise PC that combines local AI capability with an external expansion path.
If AI inference increasingly moves toward the edge, that combination could become more important than simply putting a faster processor inside a smaller chassis.
Market Landscape
The AI PC market is evolving from a feature category into an infrastructure layer for enterprise computing.
Gartner expects AI PCs to represent 54.7% of worldwide PC shipments in 2026, while its research indicates that x86 Windows systems remain particularly important in business environments.
At the same time, Gartner forecasts global AI spending of $2.59 trillion in 2026, with AI infrastructure representing more than 45% of the total.
That creates a market between traditional PCs and data-center infrastructure. Compact AI workstations can provide local inference and development capabilities without requiring every workload to run on expensive cloud GPU infrastructure.
The key competitive variables will be memory capacity, accelerator performance, software compatibility, power efficiency, expandability and security. Acer’s OCuLink implementation gives the Veriton RI110 an advantage in expandability, while NVIDIA-powered workstations can offer substantially greater dedicated GPU performance for heavier AI workloads.
For IT departments, the product is best viewed as a potential hybrid AI endpoint, rather than a replacement for enterprise GPU clusters or cloud AI platforms.
Top Insights
- Acer’s Veriton RI110 brings local AI inference to a compact workstation, giving developers and enterprise teams an alternative to sending every workload to cloud-based GPU infrastructure.
- Up to 96GB memory and support for large AI models make the system relevant to private inference, development and content workflows where local data processing matters.
- OCuLink adds an external GPU upgrade path, allowing organizations to expand AI compute without immediately replacing the compact workstation or moving workloads entirely into servers.
- Acer Qubi Claw introduces sandboxed local AI agents, highlighting the growing enterprise focus on autonomous workflows, privacy controls and local execution.
- AI PCs are becoming mainstream enterprise hardware, with Gartner forecasting that they will account for 54.7% of PC shipments worldwide in 2026.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
