Alex Jackson · October 20, 2026 
NVIDIA has officially started selling its new RTX Spark laptops, the company’s boldest push yet into AI PCs. The machines went on sale globally on October 16, 2026, and NVIDIA is pitching them as laptops that can run large AI models and local AI agents entirely on the device — no cloud connection required. Prices start at $2,599.
What Are RTX Spark Laptops?
Every RTX Spark laptop is built around a single NVIDIA “superchip” that pairs a Blackwell-based RTX GPU with a 20-core Grace CPU, an Arm design co-developed with MediaTek. The two processors sit side by side with up to 128GB of unified LPDDR5X memory shared between them, linked at 600 GB/s via NVLink-C2C.
The top configuration packs 6,144 CUDA cores, fifth-generation Tensor Cores and around 1 petaflop of FP4 AI performance. Because the CPU and GPU draw from the same memory pool, these laptops can hold 3D scenes over 90GB, edit 8K 4:2:2 video and load AI models that would choke a conventional laptop with 16 or 32GB of RAM.
Run AI Models and Agents Locally
This is the real story of RTX Spark. NVIDIA says the platform runs local AI agents and open-weight models out of the box. Agent applications such as GitHub Copilot, Perplexity, OpenClaw, Hermes, Pi and WorkBuddy get streamlined setup on RTX Spark with added security features, and they can recommend optimized local models and configurations automatically.
Alongside the hardware, NVIDIA released Nemotron 3.5 Bolt 74B — a 74-billion-parameter open-source AI model compressed to just 24GB using 2-bit quantization, designed specifically for agentic AI on RTX Spark. It works through popular local-AI tools like Ollama, LM Studio, llama.cpp, vLLM and Unsloth.
NVIDIA PAIR (Personal AI Router), a free open-source tool that spreads AI workloads across compatible PCs on a local network, also moved out of beta on October 13. Microsoft’s Yusuf Mehdi summed up the vision: Windows is becoming a platform for “hybrid intelligence, where AI can run locally when it makes sense and reach the cloud when needed.”
Models, Prices and Availability
The launch lineup spans nine models from six manufacturers: ASUS ProArt P14 and P16, Dell XPS 16, HP OmniBook Ultra and OmniBook X 14, Lenovo Yoga Pro 9n, Microsoft Surface Laptop Ultra, and the MSI Prestige N16 Flip AI+ and Prestige N16.
Prices climb quickly with memory, and that’s the catch — the $2,599 entry price buys the base configuration. The full 128GB Microsoft Surface Laptop Ultra reaches $5,899.99, while a maxed-out ASUS ProArt P16 tops out at $6,999.99. NVIDIA is also running a limited-time promotion for new Perplexity users.
Also a Gaming and Creative Powerhouse
Gamers are not left out. NVIDIA says RTX Spark machines can play AAA titles at 1440p and over 100 frames per second, backed by DLSS 5, ray tracing, G-SYNC and Reflex. Launch-day support covers Cyberpunk 2077, Fortnite, Gears of War: E-Day, Marvel Rivals, World of Warcraft and NBA 2K27 — the last of which uses DLSS 5 with 3D-Guided Neural Rendering for richer lighting and more realistic materials.
What Is Coming Next
NVIDIA says compact desktops built for always-on AI agents are scheduled for November, and the DGX Station for Windows — with 748GB of coherent memory and up to 20 petaflops of FP4 compute — is scheduled for December. More laptop models from additional partners are expected to follow.
Should You Buy One?
If you are a developer, data scientist or creator who runs AI models locally, RTX Spark is the first Windows laptop line built specifically around that workload — and the unified memory is genuinely different from anything in a normal laptop. But be realistic: independent reviews are only just starting to appear, and real-world battery life, software compatibility and sustained performance are still unverified. If you just need a great everyday laptop, the entry price is hard to justify until the local-AI ecosystem matures. For AI-first buyers, though, this launch is a milestone.
RELATED STORIES
