AI Models & Platforms
Surface RTX Spark Dev Box Ships in November at $5,999

Microsoft opened preorders for the Surface RTX Spark Dev Box on October 7, 2026, pricing the compact NVIDIA-powered developer PC at $5,999 MSRP and selling it exclusively through Microsoft.com in the United States, with shipments to customers beginning in November.
Brett Ostrum, Microsoft’s corporate vice president of Surface, announced the preorder opening in a post on the Microsoft Devices Blog, writing that the company was opening preorders for two machines built around NVIDIA’s RTX Spark superchip: the Dev Box and the Surface Laptop Ultra. He presented the pair as part of a broader Windows vision for hybrid intelligence that Microsoft outlined the same day, combining the power of the PC and the cloud. “Windows is the OS for agents, and Surface is the hardware for Windows,” Ostrum wrote.
The Dev Box is listed on Microsoft’s product page “From $5,999.99” with an active preorder link, and the page notes that an external monitor and keyboard are sold separately. The Surface Laptop Ultra starts at $2,599 MSRP from Microsoft.com and select retailers worldwide, with availability beginning October 16, 2026. Microsoft is offering up to $1,000 cash back with the trade-in of an eligible MacBook Pro on qualifying Surface Laptop Ultra purchases made between October 7, 2026, and November 23, 2026, through the Microsoft Store online in the United States and Canada.
RTX Spark Hardware and Chassis Design
Microsoft first introduced the Dev Box at its Build developer conference on June 2, 2026, in a Devices Blog post from Andrew Hill, corporate vice president of Surface Product, describing a compact developer PC that pairs an NVIDIA Blackwell RTX GPU with an NVIDIA Grace CPU. Microsoft said at the time that the system could run models exceeding 120 billion parameters with a 1 million token context window locally at interactive speeds, and committed to availability later in 2026 in the U.S. exclusively on Microsoft.com.
NVIDIA unveiled the RTX Spark platform itself on May 31, 2026, at GTC Taipei. According to NVIDIA’s announcement, the superchip combines a Blackwell RTX GPU with 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision with a 20-core Grace CPU, connected through the NVLink-C2C chip-to-chip interconnect, and MediaTek collaborated with NVIDIA on the custom CPU design.
In the October 7, 2026 announcement, Microsoft said the Dev Box delivers up to 1 petaflop of AI performance with 128 GB of unified memory and supports AI models exceeding 120 billion parameters with large context windows. Footnotes to the post state that the petaflop figure is based on theoretical FP4 performance using the sparsity feature, and that the 128 GB is dynamically shared between the CPU and GPU, with the maximum amount addressable by the GPU dependent on system configuration and workload.
The enclosure is an anodized aluminum body with 1,000 air vents built into its grid chassis, which Microsoft describes as a nod to 1,000 teraflops of compute performance. A 100W thermal envelope sits inside the aluminum chassis, which doubles as a cooling system that Microsoft says is designed to keep compute power consistent through long training runs. Ports include two USB-C, USB-A, HDMI, Ethernet, and a headphone jack.
Preconfigured Developer Software
As part of what Microsoft calls the Project Zenith family of devices, the Dev Box ships with Visual Studio Code, Git, GitHub CLI, GitHub Copilot, WSL, Python, and Node among its preinstalled tools. Windows 11 Pro is preconfigured at the image level with a dark theme, a simplified taskbar, Widgets removed, Do Not Disturb enabled, Developer Mode turned on, and PowerShell 7 as the default shell, while WSL 2 is configured with GPU passthrough and CUDA support.
Microsoft’s June post ties the machine to the company’s wider developer stack: AI Toolkit for VS Code brings model conversion, fine-tuning, and evaluation into the editor; Windows ML with TensorRT provides a local inference surface; and Microsoft Foundry connects local prototyping to production deployment.
On the product page, Microsoft states that Windows ML with the TensorRT-RTX Execution Provider optimizes GPU performance for bring-your-own-model workloads, and that a local Aion 1.0 Instruct model enables text intelligence and agent tool calling directly on the device. Microsoft also says intelligent routing in GitHub Copilot can offload up to 20 percent of workloads using MAI-1-Code-Flash and Nemotron-3-Super, and that the machine runs coding models of up to 120 billion parameters locally.
Security and Agent Controls
The Dev Box is a Windows 11 secured-core PC and works with BitLocker encryption, Microsoft Defender, Entra ID, and Intune for management and governance. Microsoft said Microsoft Execution Containers provide containment for agents running on the device, and that upcoming Windows capabilities will use Microsoft Entra to distinguish agent activity from user activity while extending Microsoft Agent 365 controls to local agents on-device. With up to 128 GB of unified memory, the company said, both new Surface machines can run demanding models and agent workflows locally, including GitHub Copilot tasks that investigate builds and propose fixes, while Microsoft Dev Box adds cloud capacity when needed.
Beyond Microsoft, NVIDIA said in its May 31, 2026 release that RTX Spark compact desktops would be available in fall 2026 from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with models from Acer and GIGABYTE to follow.












