News in short

• NVIDIA has announced global availability of RTX Spark laptops from ASUS, Dell, HP, Lenovo, Microsoft and MSI.
• The new PCs combine NVIDIA Blackwell RTX graphics and a Grace CPU, with up to 128GB of unified memory.
• NVIDIA claims up to one petaflop of FP4 AI performance for running AI models and agents locally.
• The systems target developers, creators and gamers, with support for coding agents, creative applications and RTX gaming features.
• RTX Spark laptops start at $2,599 globally, with pricing and configurations varying by region.

NVIDIA has announced the global availability of its RTX Spark PCs, a new category of Windows computers designed for local artificial intelligence, creative workloads and gaming. The systems combine NVIDIA Blackwell RTX graphics, a Grace CPU and up to 128GB of unified memory.

Announced on October 16, 2026, the lineup is available from ASUS, Dell, HP, Lenovo, Microsoft and MSI. Prices start at $2,599, although configurations and pricing vary by region.

NVIDIA says RTX Spark delivers up to one petaflop of FP4 AI performance. The company is positioning the hardware as a portable platform for running AI models and agents locally, alongside demanding creative and gaming applications.

RTX Spark brings local AI to Windows PCs

The new systems combine an NVIDIA Blackwell RTX GPU and NVIDIA Grace CPU in a single superchip. Their unified memory allows supported workloads to share system resources instead of relying on separate pools of CPU and GPU memory.

NVIDIA says the hardware can run large language models and AI agents locally. It also targets creative projects involving 8K video and 3D scenes exceeding 90GB.

MediaTek collaborated with NVIDIA on the custom CPU design. The partnership focuses on power efficiency, performance and connectivity for portable computers.

The RTX Spark platform also supports NVIDIA technologies such as CUDA, DLSS, OptiX, Reflex, G-SYNC, TensorRT and hardware-accelerated video. These capabilities bring AI development and graphics acceleration into the same Windows device.

AI agents can handle more than coding

NVIDIA is positioning RTX Spark as a platform for AI agents that can complete tasks, generate assets and write code. Guided setup and installers are designed to help users get started without relying on command-line workflows.

The supported ecosystem includes GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy. NVIDIA is also introducing support for its Nemotron 3.5 Bolt 74B-A9B 24GB model, which is optimised for tool calling and multistep tasks on supported configurations.

For security, NVIDIA highlights its OpenShell runtime and Microsoft Execution Containers. These technologies are intended to help users control agent permissions and manage how agents interact with tools.

NVIDIA PAIR can also let RTX Spark serve AI inference to compatible Windows, Linux and macOS computers on a home network. This could allow other devices to access local AI computing without running the same workloads themselves.

Creative applications get RTX acceleration

RTX Spark targets creators who need portable computing for video editing, image processing and 3D work.

Adobe is adding NVIDIA TensorRT acceleration to AI editing features in applications including Photoshop, Premiere and Substance 3D. Photoshop and Premiere have beta releases with native RTX Spark support available at launch, while Substance support is planned for later.

NVIDIA Broadcast 2.3 adds tools for removing background noise from existing recordings and shaping voices with Studio Voice Mic Profiles. Beta RTX Video features support up to 4K super resolution and four-times frame generation.

The company also highlights ComfyUI workflows that can convert AI-generated video from 15 frames per second to smoother 60fps footage. These capabilities are designed to reduce reliance on per-use cloud inference for supported tasks.

Gaming and software development on one machine

RTX Spark is also designed for gaming, with support for NVIDIA technologies such as DLSS, ray tracing, G-SYNC and Reflex. NVIDIA lists games including Cyberpunk 2077, Fortnite, Gears of War: E-Day, Marvel Rivals, World of Warcraft and NBA 2K27 among the supported titles.

The company says the platform can run AAA games at 1440p with frame rates above 100fps in supported scenarios. Actual results will depend on the game, settings and specific system configuration.

For developers, native CUDA support enables local model execution and development workflows. Windows Subsystem for Linux provides access to familiar Linux tools, while NVIDIA’s developer resources aim to simplify work with CUDA-X libraries.

NVIDIA and Microsoft have also prepared development support for Windows on Arm, including a PyTorch-CUDA build and native Python and Conda tools.

RTX Spark availability and pricing

RTX Spark laptops are available globally through ASUS, Dell, HP, Lenovo, Microsoft and MSI websites and major retailers. Starting prices are $2,599, with regional pricing and configurations varying.

The announced lineup includes the ASUS ProArt P14 and P16, Dell XPS 16, HP OmniBook Ultra and OmniBook X 14, Lenovo Yoga Pro 9n, Microsoft Surface Laptop Ultra, and MSI Prestige N16 Flip AI+ and Prestige N16.

NVIDIA has also announced a limited-time promotion for new Perplexity users. Buyers should check regional listings for eligibility, configurations and final pricing.

RTX Spark’s appeal will depend on whether users need local AI computing alongside creative and gaming performance. Its combination of unified memory, accelerated graphics and agent support targets people who want to run more workloads on a single Windows PC, rather than depending entirely on cloud services.