NVIDIA announced the RTX Spark Superchip at Computex 2026 during CEO Jensen Huang's keynote on May 31, marking the company's first entry into the PC main-processor market. The chip combines a 20-core Arm-based Grace CPU with a Blackwell RTX GPU on a Windows on Arm platform, positioning NVIDIA to compete directly with Intel, AMD, Qualcomm and Apple for the heart of consumer computing.
According to NVIDIA, the RTX Spark is designed to "reinvent Windows PCs for the era of personal AI agents," moving the PC from "tool to teammate." Partner laptops and desktops from major OEMs are slated to arrive in Fall 2026.
Key Highlights
- First PC processor from NVIDIA. The RTX Spark Superchip is NVIDIA's debut as a central PC processor, challenging the x86 (Intel/AMD) and Qualcomm Snapdragon X Windows-on-Arm establishment.
- Hybrid architecture. A 20-core Grace CPU and a Blackwell RTX GPU with 6,144 CUDA cores are linked by NVLink-C2C chip-to-chip interconnect.
- Up to 1 petaflop of AI compute (FP4) and up to 128GB of unified memory.
- Windows on Arm platform built with Microsoft, including new Windows security primitives and a runtime called NVIDIA OpenShell for Windows.
- Broad OEM support from ASUS, Dell, HP, Lenovo, Microsoft Surface and MSI in Fall 2026, with Acer and GIGABYTE following.
Details
At the core of the RTX Spark is a pairing of two NVIDIA designs. The 20-core Grace CPU, based on the Arm instruction set, handles general-purpose computing, while the Blackwell RTX GPU delivers graphics and AI acceleration through 6,144 CUDA cores and 5th-generation Tensor Cores with FP4 support.
The two are connected via NVIDIA's NVLink-C2C chip-to-chip interconnect, which NVIDIA says allows the CPU and GPU to share data with high bandwidth and low latency. Configurations offer up to 128GB of unified memory accessible to both processors.
NVIDIA states the platform can reach up to 1 petaflop of AI compute using FP4 precision, a figure aimed squarely at on-device AI workloads. DLSS is supported across the RTX Spark ecosystem, including DLSS 4.5, which NVIDIA announced at the same Computex event.
On the software side, the RTX Spark runs Windows on Arm. NVIDIA partnered with Microsoft on new Windows security primitives and introduced a runtime it calls NVIDIA OpenShell for Windows.
Impact
The RTX Spark represents the most significant challenge yet to the duopoly that has defined Windows PCs: x86 chips from Intel and AMD, and Arm-based Snapdragon X processors from Qualcomm. By delivering its own CPU rather than only a discrete GPU, NVIDIA is competing for the central processor socket for the first time.
The move also draws NVIDIA into more direct comparison with Apple, whose Arm-based silicon combines CPU and GPU with unified memory in a similar fashion. NVIDIA frames the RTX Spark around "personal AI agents," signaling a bet that on-device agentic AI, rather than raw clock speed, will define the next generation of consumer machines.
For the broader market, the chip tests whether NVIDIA's dominance in data-center AI hardware can translate into consumer PCs, where software compatibility on Windows on Arm has historically been a hurdle.
Background
NVIDIA built its business on graphics processors and, more recently, on AI accelerators that power much of the data-center boom. The RTX Spark extends that lineage by folding NVIDIA's GPU expertise together with a full Arm-based CPU into a single consumer platform.
The Grace CPU and Blackwell GPU architectures both originate in NVIDIA's data-center work, and the NVLink-C2C interconnect used in the RTX Spark is a technology NVIDIA has previously deployed to tie its server-class chips together. With the RTX Spark, NVIDIA repackages those building blocks for laptops and small desktops.
What's Next
NVIDIA says RTX Spark systems will be available in Fall 2026 from ASUS, Dell, HP, Lenovo, Microsoft Surface and MSI, with Acer and GIGABYTE joining later.
Designs will span laptops as slim as 14mm and as light as 3 pounds, in 14- and 16-inch sizes, with color-accurate tandem OLED displays and NVIDIA G-SYNC. NVIDIA also confirmed small, ultra-efficient desktop form factors built on the same platform.
The launch lineup will determine how quickly the RTX Spark can establish itself against entrenched x86 and Snapdragon X competitors as the Fall 2026 window approaches.
Source: NVIDIA