NVIDIA DGX Spark Brings Local AI Computing to a Tiny PC

The acceleration of artificial intelligence is reshaping the PC market. Major chipmakers are developing “AI PC” designs that combine Windows applications with dedicated, local AI subsystems.

Alongside those client systems, a separate class of compact computers is emerging for AI workloads. These machines are built primarily to accelerate models and agents locally, reducing the need to rely entirely on remote computing.

Last year, NVIDIA introduced DGX Spark, described as the first PC of its kind focused on large language models and Agents. NVIDIA presented the system as an “AI Supercomputer,” placing advanced AI computing in a small form factor.

The product reflects a broader shift toward dedicated hardware for local AI processing. Rather than adding AI capabilities only to conventional computers, manufacturers are also developing boxes designed specifically for model and agent workloads.

After NVIDIA introduced DGX Spark, other manufacturers began describing their own compact systems as Mini PCs for AI. The growing range of labels shows how quickly the PC segment is adapting to local acceleration for AI models and agents.