NVIDIA just dropped the biggest PC news of the year. At Computex 2026 in Taipei, CEO Jensen Huang unveiled the RTX Spark superchip, a brand new processor designed with Microsoft to turn Windows laptops and desktops into proper AI machines.
Think of RTX Spark as NVIDIA officially walking into the PC business. It is a single superchip that combines an ARM CPU, a Blackwell RTX GPU, and 1 petaflop of AI performance, all aimed at running AI agents, creator tools, and AAA games on the same laptop.
Here is the short version. RTX Spark laptops and small desktops arrive this fall from Surface, Dell, HP, Lenovo, ASUS, and MSI. They can run 120 billion parameter AI models locally, edit 12K video, and play AAA games at 1440p over 100 FPS. Adobe is rebuilding Photoshop and Premiere for it. This is NVIDIA’s challenge to Apple silicon and Qualcomm’s Snapdragon X, and it is a big one.
1. What Is NVIDIA RTX Spark?
RTX Spark is a new superchip from NVIDIA built specifically for Windows PCs. It was officially unveiled at NVIDIA GTC Taipei on May 31, 2026, after almost a year of rumours under the codenames N1 and N1X.
Three things make it a big deal. First, it is NVIDIA’s first chip designed for personal computers rather than data centres or gaming graphics cards. Second, it is ARM-based, which lines it up against Apple silicon and Qualcomm’s Snapdragon X. Third, it has been built hand-in-hand with Microsoft so Windows can run AI agents safely on the device itself.
On stage, Jensen Huang summed up the pitch: for forty years, you launched apps on a PC by clicking and typing. With RTX Spark, you ask, and the PC does the work. The chip is the hardware side of that promise.
2. RTX Spark Specs at a Glance
Here is the chip on a single page.
| Component | Detail |
| GPU | NVIDIA Blackwell RTX with 6,144 CUDA cores |
| Tensor Cores | 5th generation, FP4 precision |
| CPU | 20-core NVIDIA Grace, ARM-based, co-designed with MediaTek |
| Interconnect | NVIDIA NVLink C2C between CPU and GPU |
| Unified memory | Up to 128GB LPDDR5X |
| AI performance | 1 petaflop (1,000 trillion operations per second) |
| Process node | TSMC 3nm |
| Local AI model size | Up to 120 billion parameters, 1 million tokens context |
| Form factors | Slim 14-16 inch laptops and compact desktops |
| Software stack | CUDA, RTX, DLSS 4.5, FP4, TensorRT, OptiX, Reflex, G-SYNC |
3. What RTX Spark Can Actually Do
Specs are great, but the real question is what you can do with this thing. Three areas matter.
Run powerful AI locally
This is the headline use case. RTX Spark can run large language models with up to 120 billion parameters and 1 million tokens of context, entirely on your device, with no internet round trip and no data leaving your laptop. That is data-centre-class AI compressed into something you can carry in a backpack.
Combined with the new Windows agent runtime, this means the next wave of personal agentic AI assistants can live on your PC instead of in the cloud, which has real implications for privacy, speed, and offline use.
Edit and generate professional video
RTX Spark laptops can edit 12K video in 4:2:2 colour, generate 4K AI video, and render 3D scenes up to 90GB in size with OptiX path tracing and DLSS. That is the kind of work that used to need a tower workstation with multiple discrete GPUs.
Play AAA games on a slim laptop
On the gaming side, RTX Spark hits AAA titles at 1440p over 100 frames per second with ray tracing, DLSS, and Reflex enabled. NVIDIA demoed Forza Horizon 6 and 007 First Light on stage. It also introduces DLSS 4.5 Ray Reconstruction with a second-generation transformer model, coming to Blender 5.3 and dozens of games.
4. The NVIDIA and Microsoft Partnership: AI Agents on Windows
The chip is half of the story. The other half is what Microsoft is doing inside Windows itself to make on-device AI agents safe and useful.
New Windows security primitives
Microsoft is adding new system-level building blocks to Windows that handle identity, containment, policy, and end-to-end security for AI agents. The short version: agents will get their own permissions model, similar to how apps already do, so they cannot freely roam through your files or your accounts without consent.
NVIDIA OpenShell
Layered on top is NVIDIA OpenShell, a new runtime that decides what an agent is allowed to do, intelligently routes queries to local models for sensitive tasks, and disguises personal information before sending anything to a cloud model. The aim is that you keep the convenience of AI assistants without giving up control of your data.
Independent agent makers including OpenClaw and the Nous Research Hermes Agent have already signed up to ship Windows apps that use this stack. Expect a wave of new agent apps to land on the Windows taskbar alongside the first RTX Spark laptops this fall.
Worth noting: Microsoft Build runs June 2 to 3, 2026, with more details on Windows agent capabilities and the new security primitives for developers. If you build software, that is the next event to watch.
5. Which Laptops and Desktops Will Have RTX Spark?
Eight hardware makers are launching RTX Spark systems this fall, with NVIDIA forecasting around 30 laptops and 10 desktops in total over the next year. The first wave is impressive.
Microsoft Surface Laptop Ultra
Microsoft’s flagship Surface gets a major upgrade. As detailed on the Windows Experience Blog, the new Surface Laptop Ultra is the most powerful Surface ever built, pairing Blackwell RTX with up to 128GB of unified memory for sustained creative and AI workloads.
Dell XPS 16 Creator Edition
Dell is shipping the XPS 16 Creator Edition with RTX Spark inside, aimed at people who need the full memory ceiling for video editing, 3D work, and local AI development. Alienware gaming laptops with the same chip are expected to follow.
HP OmniBook
HP is positioning its upcoming OmniBook lineup as one of the thinnest RTX Spark laptops, marketed primarily at agentic developers and creators who want portability plus power.
ASUS, Lenovo, and MSI
ASUS is committing to the platform across its premium lineup. Lenovo will release RTX Spark systems across its global brands, leaning on its scale. MSI is positioning RTX Spark as the heart of its next compact, efficient laptops and desktops.
Acer and GIGABYTE
Both will follow with their own RTX Spark designs shortly after the first wave. Acer is expected to focus on consumer and education laptops; GIGABYTE on enthusiast creator and gamer systems.
Design specs
Across the lineup, expect slim 14 to 16 inch laptops as thin as 14mm and as light as three pounds, with precision-machined aluminium chassis, color-accurate tandem OLED displays with G-SYNC, and all-day battery life. NVIDIA is positioning these as premium machines, so prices will sit at the high end.
6. Adobe and Gaming Get Big Upgrades
Adobe rebuilds Photoshop and Premiere
This is one of the more surprising parts of the announcement. Adobe is rearchitecting Photoshop and Premiere from the ground up for RTX Spark, with up to 2x faster AI and graphics performance compared to current versions.
Premiere gets a new video pipeline that taps RTX Spark’s unified memory and TensorRT software, delivering real-time editing and colour correction with GPU-accelerated AI. Photoshop’s next-gen engine handles GPU-accelerated compositing, live filters, HDR, and modern natural brushing. Substance 3D Painter and Stager are getting native versions too.
Both apps will also support Windows agent workflows, so you can have an AI agent help with edits inside the creative app itself, instead of switching back to a chatbot tab. Updates begin rolling out alongside RTX Spark hardware this fall.
RTX Spark in over 1,000 games and apps
The full NVIDIA RTX software stack supports more than 1,000 games and apps. On the games side, KRAFTON, NetEase, Remedy Entertainment, Riot Games, and even XBOX on PC are embracing the platform. New DLSS 4.5 Ray Reconstruction lands in Blender 5.3 and dozens of games this fall, and ComfyUI gets RTX Video with 4x Frame Generation.
7. When Can You Buy One, and What Will It Cost?
RTX Spark laptops and compact desktops arrive this fall (autumn 2026 in the northern hemisphere) from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE models following shortly after.
NVIDIA has not announced pricing. Industry coverage is treating these as premium machines, in the same bracket as high-end MacBook Pro and top-tier creator and gaming laptops. Expect the lower configurations to land around the same price as today’s flagship Windows ultrabooks and the maxed-out 128GB versions to sit well above that.
If you want maximum power for AI development, video editing, or running large language models locally, this is going to be the most exciting Windows laptop generation in years. If your needs are mostly browser, email, and Office documents, current Intel and AMD laptops still do that job for a fraction of the price.
8. Why This Matters: NVIDIA vs Apple vs Qualcomm
Zoom out for a second. RTX Spark is more than a new chip. It changes the shape of the PC market.
NVIDIA officially enters the PC business
Until now, NVIDIA sold GPUs that went into PCs. RTX Spark is the first time NVIDIA has shipped the brain of the PC itself, the chip that combines CPU, GPU, NPU, and memory in one package. That is a huge strategic shift, and it gives NVIDIA a direct relationship with PC buyers, not just with PC builders.
A real challenge to Apple silicon
Apple’s M-series chips are the gold standard for premium laptops right now: efficient, powerful, beautifully integrated. RTX Spark is NVIDIA’s answer. The combination of a 20-core ARM CPU, a 6,144-core Blackwell GPU, and 128GB of unified memory is the first credible challenge to MacBook Pro on its own terms, particularly for AI and creative work.
Pressure on Qualcomm Snapdragon X
Qualcomm has spent the last two years pushing Snapdragon X chips into Windows laptops, championing ARM-on-Windows. RTX Spark is a far more powerful ARM chip and arrives with deeper Windows integration thanks to the NVIDIA and Microsoft partnership. Qualcomm is now the budget ARM option, not the premium one.
Annual updates
NVIDIA also confirmed that RTX Spark will get annual updates alongside the rest of its AI stack. So this is not a one-off chip; it is the start of a new product line. That cadence matches Apple’s roughly annual M-chip cycle and signals NVIDIA is in this for the long haul.
For broader AI context on the Windows side, our beginner’s guide to Google AI Mode covers how AI is changing Search at the same time. RTX Spark will let those AI experiences run far more capably on your own device.
Frequently Asked Questions
What is NVIDIA RTX Spark?
RTX Spark is a new superchip from NVIDIA designed for Windows laptops and compact desktops. It combines a 20-core ARM CPU, a Blackwell RTX GPU with 6,144 CUDA cores, an NPU, and up to 128GB of unified memory on a single TSMC 3nm package, delivering 1 petaflop of AI performance.
When will RTX Spark laptops be available?
This fall, 2026, from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE models to follow.
How much will RTX Spark laptops cost?
NVIDIA has not announced pricing yet. Industry coverage expects premium positioning, in the same price bracket as flagship MacBook Pro and top-tier creator and gaming Windows laptops.
Is RTX Spark a CPU or a GPU?
Both. RTX Spark is a single superchip that includes a 20-core NVIDIA Grace ARM CPU, a Blackwell RTX GPU, and an NPU, all connected by NVLink C2C and sharing up to 128GB of unified memory.
Can RTX Spark run large language models locally?
Yes. NVIDIA states that RTX Spark can run LLMs with up to 120 billion parameters and 1 million tokens of context, all on the device, without sending queries to the cloud.
Is RTX Spark only for AI, or also for gaming?
Both. NVIDIA designed RTX Spark for AI, content creation, and gaming. It can play AAA titles at 1440p over 100 frames per second with ray tracing, DLSS, and Reflex enabled, and the full NVIDIA RTX software stack including DLSS 4.5 Ray Reconstruction supports over 1,000 games and apps.
How is RTX Spark different from Apple silicon and Qualcomm Snapdragon X?
All three are ARM-based chips, but RTX Spark is the most powerful by a clear margin in early specs, especially for AI workloads. It also brings the full NVIDIA CUDA, RTX, and DLSS software stack to Windows, which neither Apple silicon nor Snapdragon X offers.
What is NVIDIA OpenShell?
OpenShell is a new runtime from NVIDIA that controls what AI agents can do on your PC, routes sensitive queries to local models, and disguises personal information before sending anything to a cloud model. It works alongside new Microsoft Windows security primitives to keep on-device agents safe.
If you found this useful, you may also like our guide to agentic AI, which covers the broader trend that RTX Spark is built to power.
