Nvidia RTX Spark Superchip Targets Windows PCs

Nvidia is no longer content to sell the graphics card inside your PC. At Computex, Jensen Huang unveiled RTX Spark, a superchip that bundles an Arm CPU, a Blackwell GPU, and 128GB of unified memory, and it is aimed squarely at running serious AI models on a Windows machine.

Key takeaways

• RTX Spark is a full superchip, Arm CPU plus Blackwell GPU on one package.
• 128GB of unified memory lets the CPU and GPU share the same pool.
• Up to 20 CPU cores and 6,144 CUDA cores with up to 300 GB/s bandwidth.
• Hardware ships in the fall, with over 30 laptops and around 10 desktops planned.
• Rivals felt it immediately, as AMD, Intel, and Qualcomm shares slipped on the news.

What makes RTX Spark different

The headline feature is unified memory. On a conventional PC, the CPU and GPU each have their own memory, and moving data between them is a bottleneck for AI work. RTX Spark puts the CPU and GPU on a single system-on-chip that shares one 128GB pool, eliminating that copy step. The practical result is the ability to load and run bigger, more capable AI models locally instead of shipping every request to the cloud.

The specifications back up the ambition: up to 20 Arm CPU cores, a Blackwell GPU with 6,144 CUDA cores, 128GB of LPDDR5X, and up to 300 GB/s of memory bandwidth. Nvidia frames this as moving the PC from a tool to a teammate, with on-device AI agents that can act rather than just answer.

Component	RTX Spark spec
CPU	Up to 20 Arm cores
GPU	Blackwell, 6,144 CUDA cores
Memory	128GB LPDDR5X, unified
Bandwidth	Up to 300 GB/s
Availability	Fall 2026

Why this rattles AMD, Intel, and Qualcomm

Nvidia entering the PC chip market is a direct threat to the companies that own it today. When Huang announced RTX Spark, shares of AMD, Intel, and Qualcomm fell as Wall Street recognized that the most valuable chip company on earth now wants a share of the laptop and desktop processor market too. This is Huang's bid to own every layer of the AI stack, from data-center accelerators down to the chip in a consumer notebook.

The competitive pressure is real because Nvidia brings its CUDA software ecosystem with it. Developers already build AI on CUDA, so a PC chip that runs the same stack locally has a built-in advantage that raw specs alone would not explain. For context on how Nvidia is pushing into new markets, our piece on Amazon's Trainium direct chip sales shows the counter-moves rivals are making.

The Microsoft partnership

RTX Spark is not just silicon. Nvidia announced it alongside Microsoft, positioning Windows as an agentic AI operating system where on-device agents handle tasks. That software partnership matters because unified memory and a fast GPU are only useful if the operating system and apps know how to feed them. Tying the chip to Windows gives it a clear home and a distribution path through major PC makers.

Those makers are already lined up. RTX Spark laptops and compact desktops will arrive in the fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and Gigabyte to follow. Nvidia expects more than 30 laptops and around 10 desktops at launch.

Element	Detail
OS partner	Microsoft, agentic Windows
Launch partners	ASUS, Dell, HP, Lenovo, Surface, MSI
Following	Acer, Gigabyte
Launch volume	30-plus laptops, ~10 desktops

What this means

If you are shopping for a laptop, RTX Spark machines could be worth waiting for if you care about running AI models locally, for privacy, latency, or working offline. The unified-memory design is genuinely suited to that, more so than a standard laptop with a discrete GPU. Just remember these are fall products, so nothing is buyable today.

For the industry, this is a shot across the bow of the established PC chipmakers. Nvidia has the AI mindshare, the software ecosystem, and now a compelling hardware platform. Whether it captures meaningful PC market share depends on price, battery life, and app support, all of which we will only learn when real machines ship. For a sense of the current landscape, see our best handheld gaming PC buying guide.

Frequently asked questions

What is Nvidia RTX Spark?

RTX Spark is a superchip Nvidia unveiled at Computex 2026 for Windows PCs. It combines up to 20 Arm CPU cores, a Blackwell GPU with 6,144 CUDA cores, and 128GB of unified memory on a single package to run AI models locally.

When can I buy an RTX Spark laptop?

Nvidia says RTX Spark laptops and compact desktops arrive in the fall of 2026 from makers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and Gigabyte to follow.

Why is unified memory a big deal?

It lets the CPU and GPU share one 128GB memory pool instead of copying data between separate pools. That removes a major bottleneck and allows the chip to load and run larger, more capable AI models on the device.

Why did AMD, Intel, and Qualcomm stock fall?

Nvidia's entry into the PC chip market threatens the companies that dominate it today. Investors sold their shares on the news, recognizing that Nvidia now competes across the full AI stack, including consumer PCs.