Microsoft Opens $5,999 Surface RTX Spark Dev Box Preorders for Local AI Development

Microsoft has opened preorders for the Surface RTX Spark Dev Box at $5,999, positioning the compact Windows workstation as a dedicated machine for developers building and testing AI systems locally. The device is scheduled to begin shipping in November, according to Microsoft’s launch materials and event coverage.

The Dev Box arrives alongside the Surface Laptop Ultra, which uses the same Nvidia RTX Spark platform in a portable design. Microsoft introduced the desktop at its October 7 Surface event after first previewing it during Build in June, linking the product to a broader push toward local model development, AI agents and Windows-based software creation.

The Surface RTX Spark Dev Box combines Nvidia’s RTX Spark processor with 128GB of memory and up to 1 petaflop of AI compute, according to product descriptions from Microsoft, Nvidia and GSM Arena. Microsoft and its partners say the system can run AI models with more than 120 billion parameters locally, a capacity aimed at workloads that typically rely on rented cloud GPUs.

That specification matters because memory often determines which models developers can load and how much data they can keep available during experimentation. Nvidia describes RTX Spark as a platform for AI development, content creation, gaming and continuously running agents, while Microsoft frames the Dev Box around longer jobs, larger models and local prototyping. The hardware remains a high-end specialist workstation rather than a general-purpose mini PC.

At $5,999, the Dev Box sits far above mainstream desktops and even above many high-end gaming systems. Its price reflects a product aimed at professional developers, research teams, startups and enterprise groups that regularly pay for cloud GPU time or need sensitive data to remain on local hardware.

The financial case depends on workload. A team running repeated inference, model tuning or agent tests can avoid some cloud charges and reduce waiting between experiments, but the purchase also shifts costs to hardware, electricity, maintenance and eventual upgrades. The Gadget Flow review emphasizes the value of the 128GB memory pool over the headline petaflop figure, a useful distinction for developers whose bottleneck is model capacity rather than peak compute.

The Dev Box is part of a wider Microsoft and Nvidia effort to make Windows a place where AI agents and large models run directly on personal computers. Jensen Huang and Satya Nadella presented the RTX Spark partnership as a new phase for Windows PCs, while Nvidia’s product materials describe compact desktops that can operate agents around the clock.

Microsoft’s Build announcements supplied the software context, including agent-focused tools across GitHub Copilot, Foundry and Copilot Studio. The company’s device announcements also describe PCs built for AI acceleration and business workloads. Together, those moves position the Dev Box as infrastructure for local development, not simply a faster Surface-branded computer.

The strongest audience is developers who work with large models, fine-tuning pipelines, autonomous agents or repeated local inference. The machine also offers a controlled environment for teams handling proprietary code or data, provided their software stack supports the RTX Spark platform and Windows.

Smaller teams and individual developers face a harder calculation. Cloud services remain easier to scale, while a $5,999 workstation concentrates capacity in one location and ties performance to a fixed hardware purchase. The Surface Laptop Ultra offers a more mobile version of the same platform, but its lower starting price reflects a different configuration and use case.

Preorders are open now, with shipping planned for November. The next practical tests will be software compatibility, sustained performance, thermals, noise, model support and how the system handles developer tools beyond the launch demonstrations.

The product also gives Microsoft and Nvidia a visible test of demand for local AI hardware. If developers adopt compact systems with workstation-class memory, other PC makers will have a clearer market for similar products. Nvidia is also promoting broader Windows AI infrastructure, including a DGX Station for enterprise desktops, indicating that the Dev Box sits within a larger hardware push rather than a one-off Surface experiment.