NVIDIA RTX Spark laptops are available globally from 16 October 2026 from ASUS, Dell, HP, Lenovo, Microsoft and MSI, through their websites and major retailers, starting at $2,599, NVIDIA said in a press release the same day. It says pricing and configurations vary by region.
The laptops are built on a single superchip that NVIDIA says pairs a Blackwell RTX GPU with a Grace CPU, with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance, enough, it says, to run large language models and AI agents locally.
Business Pill · ON-DEVICE AI
A one-minute explainer of on-device AI: running a model on the phone or PC itself instead of in a data center, and what that changes for cost, speed and privacy. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, the release puts a price on NVIDIAโs local-AI Windows laptop: RTX Spark models from six PC makers start at $2,599, sold on up to 128GB of unified memory to run large language models and AI agents locally.
The release lists the launch lineup as 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. By our count that is nine models from six makers, and NVIDIA says additional systems are to follow.
NVIDIA gives one starting price for the lineup, $2,599, and mentions “a limited-time promotion for new Perplexity users.” On 7 October we reported that Microsoft priced the Surface Laptop Ultra at $2,599 MSRP, with availability from 16 October.
“RTX Spark brings the power of NVIDIA AI, RTX graphics and CUDA together with Microsoft Windows in one PC for creating, gaming and running personal AI agents,” said Jensen Huang, founder and CEO of NVIDIA, in the release.
Yusuf Mehdi, executive vice president at Microsoft, said in the release: “We’re building Windows to be the platform for hybrid intelligence, where AI can run locally when it makes sense and reach the cloud when needed.”
NVIDIA says RTX Spark laptops are available globally from 16 October 2026 from ASUS, Dell, HP, Lenovo, Microsoft and MSI, starting at $2,599, with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. It says pricing and configurations vary by region.
What Was Posted on X
PSA: Can confirm that the NVIDIA app is automatically updating and has *enabled by default* a setting which allows AI agents to control the app with very permissive rights.
Besides the obnoxious privacy invasion and lame AI features, it’s a high security risk you may not want.
Agustin Cordes posted on 15 October that the NVIDIA app was updating automatically and had enabled by default a setting that lets AI agents control the app, calling it a privacy invasion and “a high security risk you may not want.” The security-risk view is his; the post had about 27,700 likes when we fetched it.
What Is Inside
NVIDIA says the superchip combines an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU, and that MediaTek collaborated with NVIDIA on the custom CPU design. It describes slim designs, all-day battery life and a device that “fits in a messenger bag.”
The release says creators can edit 8K 4:2:2 video and render large 3D scenes up to 90GB, even when unplugged. Adobe’s Photoshop and Premiere offer native RTX Spark support through beta releases available at launch, with Substance coming soon; we covered the Adobe and NVIDIA collaboration on 15 October.
For gaming, NVIDIA lists launch support for titles including Cyberpunk 2077, Fortnite, Gears of War: E-Day, Marvel Rivals, World of Warcraft and NBA 2K27, and says a new 2x faster AI model, available today, boosts DLSS 5 frame rates by up to 1.4x.
Agents on the Laptop
NVIDIA says guided setup and installers help users get started with agents including GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy, “without requiring a command line.” It names Nemotron 3.5 Bolt 74B-A9B 24GB, a 74-billion-parameter model optimized for tool calling and multistep tasks, for supported RTX Spark configurations.
The NVIDIA OpenShell secure runtime and Microsoft Execution Containers help users control agent permissions, the release says. NVIDIA PAIR lets RTX Spark serve AI inference to compatible Windows, Linux and macOS computers on a home network.
In ComfyUI, NVIDIA says new RTX Video nodes can turn 15 fps AI-generated video into 60 fps footage and convert it to high dynamic range “without requiring per-use cloud inference charges.”
For developers, NVIDIA says native CUDA support keeps locally processed code and context on the device, “with a path to deploy across more than 150 million NVIDIA RTX™ AI PCs and NVIDIA GPUs in the cloud.” It lists the first PyTorch-CUDA build for Windows on Arm and support for 97% of the top 15,000 Python packages.
“We brought Hermes to RTX Spark so people can put open AI models and agents to work locally on a Windows laptop, with greater control over their data and the freedom to shape AI around their own workflows,” said Dillon Rolnick, CEO at Nous Research, creators of Hermes Agent, in the release.
Update, 17 October: The NVIDIA App Turned On an AI-Agent Server by Default
The local-agent pitch behind these laptops reached NVIDIA’s existing PC software in the same week. NVIDIA App 11.0.9.565 adds a local Model Context Protocol (MCP) server that lets compatible AI assistants control app features, and VideoCardz reported on 16 October that the setting, “Allow AI agents to control NVIDIA App”, was switched on after updating without users activating it. VideoCardz said NVIDIA mentions MCP support in its release notes “but does not explain why the server is enabled by default.”
NVIDIA’s own tool reference lists what agents can call, including listing detected games, launching an application, applying game optimization presets, reading driver status and setting Battery Boost or Whisper Mode on laptops. VideoCardz says the server listens only on the local loopback address and does not connect an external AI service on its own, and that switching off the toggle under Settings, Features, Connect Agents & Apps stops it. In the post embedded below, user Agustin Cordes called the default “a high security risk you may not want.”
The Structural Read
As we read it, the release sells the machine on agents as much as on graphics: it names agents including GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy, a 74-billion-parameter Nemotron model for supported configurations, and permission controls through OpenShell and Microsoft Execution Containers.
As we read it, part of the pitch is cost: in ComfyUI, NVIDIA says the new RTX Video nodes work “without requiring per-use cloud inference charges,” and NVIDIA PAIR lets the laptop serve AI inference to other computers on a home network.
As we read it, the $2,599 starting price NVIDIA gives for the lineup is the same figure Microsoft gave for the Surface Laptop Ultra when we reported its pricing on 7 October; the release we read does not list the other makers’ prices model by model.
Yusuf Mehdi, executive vice president at Microsoft, in NVIDIA’s release, 16 October 2026
“Weโre building Windows to be the platform for hybrid intelligence, where AI can run locally when it makes sense and reach the cloud when needed”
Three Implications
BUYERS NVIDIA says the laptops are on sale from 16 October through the six makers’ websites and major retailers, starting at $2,599, and that pricing and configurations vary by region.
DEVELOPERS NVIDIA says native CUDA support keeps locally processed code and context on the device, and it lists the first PyTorch-CUDA build for Windows on Arm and support for 97% of the top 15,000 Python packages.
AGENT MAKERS The release names guided setup and installers for agents including Hermes and OpenClaw, “without requiring a command line,” and says NVIDIA agent skills let compatible agents work with tools such as NVIDIA Broadcast.
The framework argues: “Training was a cost center: episodic, concentrated among roughly twenty frontier model labs, amortizable across trillions of tokens. Inference is a revenue engine: continuous, distributed across millions of applications and agents, incurring marginal cost with every query.”
As we read it, RTX Spark is NVIDIA selling some of that inference as hardware the buyer owns: the release pitches local agents and models, and in one case work done “without requiring per-use cloud inference charges.” The framework predates the release and does not discuss RTX Spark.
The RTX Spark laptop models NVIDIA’s release names at launch, by maker: nine models from six PC makers, by our count from the release’s lineup list. NVIDIA says additional systems are to follow and that pricing and configurations vary by region.
What Is Not Established
The release we read gives a single starting price for the whole lineup; we did not check each maker’s own price list, so the price of each model and configuration is not established here. NVIDIA says pricing and configurations vary by region.
We read the NVIDIA release in full. We did not read the makers’ own product pages or the Perplexity promotion terms, and we did not contact NVIDIA, Microsoft or the PC makers.
Business Engineer Framework
The State of The Inference Economy
A Business Engineer framework on why inference, not training, is now where AI money is made, and where it runs.
NVIDIA says RTX Spark laptops are available globally from 16 October 2026 from ASUS, Dell, HP, Lenovo, Microsoft and MSI, starting at $2,599, with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance, and with additional systems to follow.
The release pitches the machines for local AI agents as well as creation and gaming, naming agents such as Hermes and OpenClaw and a 74-billion-parameter Nemotron model for supported configurations.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. We read in full NVIDIA’s press release of 16 October 2026, “NVIDIA RTX Spark PCs Available Today”; every quotation and figure comes from it, except the Surface Laptop Ultra price and availability details, which come from our 7 October report on Microsoft’s Windows blog post. We did not read the PC makers’ own product pages or the Perplexity promotion terms, and we did not contact NVIDIA, Microsoft or the PC makers. The count of nine models is ours, from the release’s lineup list.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
