June 1, 2026 · Computex Taipei

What is NVIDIA RTX Spark?

RTX Spark is NVIDIA's new Arm-based system-on-chip that brings full RTX gaming to Windows on ARM. Announced at Computex 2026, it's NVIDIA's first direct entry into the PC CPU market — and it might just change everything.

Sources: NVIDIA Newsroom, The Verge, CNBC

At a Glance

Architecture
Arm
Windows on ARM
GPU Cores
6,144
CUDA Cores
Graphics
RTX 5070
Class GPU
AI
DLSS 4
Multi Frame Gen

The Story So Far

On June 1, 2026, at Computex in Taipei, NVIDIA dropped a bombshell: the company best known for graphics cards is entering the PC CPU market. RTX Spark is a system-on-chip (SoC) that combines an Arm-based CPU with an RTX 5070-class GPU on a single die. It's designed for Windows on ARM laptops and desktops, and it brings full RTX gaming — DLSS 4, ray tracing, the works — to a platform that's historically struggled with gaming performance.

This isn't NVIDIA's first Arm chip. The company's Grace Hopper and Grace Blackwell superchips already power data centers. And the DGX Spark (formerly Project DIGITS) — a desktop AI supercomputer powered by the GB10 Grace Blackwell Superchip — was announced in March 2025 at $3,999. But RTX Spark marks NVIDIA's first consumer PC chip, putting it in direct competition with Intel, AMD, and Qualcomm.

NVIDIA CEO Jensen Huang described RTX Spark as "the most efficient PC chip ever built" and said it would "attack a lot of different price points." Microsoft is the lead launch partner, with the Surface Laptop Ultra serving as the flagship showcase device. Dell, HP, Lenovo, and ASUS are also confirmed as OEM partners.


Technical Deep Dive

Architecture

RTX Spark uses the Arm instruction set architecture (ISA), the same family used by Apple's M-series chips and Qualcomm's Snapdragon X. Arm chips are known for superior power efficiency compared to x86 (Intel/AMD), which means better battery life in laptops. But unlike previous Windows on ARM attempts, RTX Spark doesn't compromise on GPU performance.

The chip features 6,144 CUDA cores — the exact same count as the desktop RTX 5070. That means real-time ray tracing, DLSS 4 with Multi Frame Generation, and full DirectX 12 Ultimate support. Early demonstrations showed Cyberpunk 2077 running on RTX Spark hardware using DLSS 4.

Specifications (Preliminary)

Component Specification Notes
CPU Architecture Arm (custom NVIDIA cores) Based on Grace architecture
GPU RTX 5070-class (6,144 CUDA cores) DLSS 4, ray tracing, DX12 Ultimate
Memory Unified LPDDR5X (TBA capacity) Shared CPU/GPU memory
Process Node TBA Expected TSMC 3nm-class
AI Engine Integrated NPU Local AI agent acceleration
OS Support Windows on ARM Native x86 emulation included

* Specifications based on Computex 2026 announcements and partner briefings. Subject to change.

AI Capabilities

NVIDIA is positioning RTX Spark as the hardware foundation for "Personal AI" — local AI agents that run on your device rather than in the cloud. The integrated NPU (Neural Processing Unit) handles AI workloads alongside the GPU's tensor cores, enabling features like local LLM inference, real-time translation, and AI-powered content creation.

Microsoft is co-developing the software stack, with Windows 11's AI features (Copilot, Recall, etc.) optimized for RTX Spark's architecture. NVIDIA also announced that its AI agent frameworks will run natively on RTX Spark hardware.


OEM Partners & Devices

OEM Model Status Expected
Microsoft Surface Laptop Ultra Announced H2 2026
Dell Pro Max (expected) Expected H2 2026
HP TBA Expected TBA
Lenovo TBA Expected TBA
ASUS TBA Expected TBA

Additional partners (Acer, Samsung, MSI, Gigabyte) are expected to announce RTX Spark devices in the coming months. View all OEMs →


Release Timeline

June 1, 2026
RTX Spark Announced at Computex
NVIDIA and Microsoft unveil the platform. Surface Laptop Ultra shown.
H2 2026 (Expected)
US Launch
First RTX Spark devices expected to ship in the United States.
Late 2026 (Expected)
Canada, UK, EU Launch
Wave 2 rollout to major Western markets.
Early 2027 (Expected)
Asia-Pacific & Global Availability
Japan, Australia, South Korea, and remaining markets.

Note: NVIDIA's DGX Spark was announced March 2025 with a planned July 2025 launch, but faced production delays. RTX Spark PC may follow a similar pattern. We update this timeline as new information becomes available.


Frequently Asked Questions

Is RTX Spark the same as DGX Spark?

No. DGX Spark (formerly Project DIGITS) is a desktop AI supercomputer powered by the GB10 Grace Blackwell Superchip, priced at $3,999. RTX Spark is a consumer PC chip for laptops and desktops, aimed at mainstream price points. Both use Arm architecture but serve different markets.

Will RTX Spark run all my Windows apps?

Windows on ARM includes x86/x64 emulation, so most Windows applications should run. Gaming compatibility depends on anti-cheat software and driver support. NVIDIA's driver stack is expected to be more mature than Qualcomm's, given the company's decades of GPU driver experience.

How much will RTX Spark PCs cost?

NVIDIA says RTX Spark will target multiple price points. Based on current Windows on ARM laptop pricing (Snapdragon X starts at ~$999) and the integrated RTX 5070-class GPU, expect RTX Spark laptops to range from $1,200 (entry) to $2,500+ (premium). Official pricing TBA.

Should I wait for RTX Spark or buy now?

If you need a laptop immediately and gaming is important, current RTX 40-series or RTX 50-series laptops are available now. If you can wait 6-12 months and want cutting-edge efficiency with RTX gaming, RTX Spark is worth the wait. Use our country tracker to monitor availability in your region.


Track RTX Spark in Your Country

Select your country to see release status, available models, and get notified when RTX Spark launches near you.