Nvidia RTX Spark Superchip: Arm-Based SoC With RTX 5070-Class GPU, 1 Petaflop AI, and 1440p Gaming at 100+ FPS

Nvidia officially unveiled the RTX Spark superchip at Computex 2026 on May 31, 2026 . The system-on-chip fuses a 20-core Nvidia Grace Arm CPU with a Blackwell RTX GPU and up to 128 GB of unified memory into a single package, targeting Windows on Arm laptops and compact desktop PCs . It is Nvidia's first consumer-class superchip, designed to bring workstation-class AI and RTX 5070-level gaming into thin, power-efficient devices .

Specifications: N1X Flagship Configuration

The flagship RTX Spark configuration, codenamed N1X, includes the following :

• CPU: 20 Arm cores (10 Cortex-X925 performance cores at 4.1 GHz + 10 Cortex-A725 efficiency cores)
• GPU: Blackwell RTX GPU with 6,144 CUDA cores (matching the desktop RTX 5070's core count)
• AI Compute: Up to 1 petaflop at FP4 precision via 5th-gen Tensor Cores
• Memory: Up to 128 GB unified LPDDR5X, ~300 GB/s bandwidth
• TDP (laptop): Estimated 45–80 W for N1X; lower for the standard N1 chip
• Manufacturing: TSMC 3 nm process, co-developed with MediaTek, using NVLink-C2C chip-to-chip interconnect

Gaming, AI, and Creative Performance

Nvidia claims RTX Spark can play AAA games at 1440p resolution with ray tracing at over 100 fps . The company demonstrated this at Computex with titles such as Forza Horizon 6 and Indiana Jones and the Great Circle . The chip supports the full Nvidia gaming stack: DLSS 4.5 (including Ray Reconstruction), RTX ray tracing, Reflex, and G-SYNC .

For AI and creative workloads, RTX Spark runs the complete CUDA, TensorRT, OptiX, and FP4 acceleration stack, capable of running 120-billion-parameter large language models locally with up to 1 million tokens of context . Nvidia says the chip can render ultralarge 90 GB 3D scenes, edit 12K 4:2:2 video, and generate 4K AI videos entirely on-device .

How It Compares to x86 Gaming PCs

RTX Spark delivers RTX 5070-class GPU performance at a fraction of the power: 45–80 W versus 150 W+ for a desktop RTX 5070 . This enables thin-and-light laptops as slim as 14 mm and mini PCs that were previously unable to deliver this level of gaming or AI performance . However, several trade-offs separate it from traditional x86 rigs:

• Memory bandwidth: A desktop RTX 5070 uses dedicated GDDR7 VRAM with approximately 672 GB/s bandwidth, while RTX Spark shares unified LPDDR5X at 273–300 GB/s .
• Game compatibility: x86 runs the entire Windows game library natively; Arm may require emulation for some x86-only legacy titles .
• Upgradeability: Desktop x86 PCs allow component swaps; RTX Spark's unified memory and SoC design are fixed .
• CPU benchmarks: Pre-release Geekbench 6 scores from an HP prototype place the N1X at approximately 3,096 single-core and 18,837 multi-core, roughly equivalent to Intel's Core Ultra 9 275HX or AMD's Ryzen AI MAX+ 395, trailing top-tier x86 by about 10–15% in multi-threaded workloads .

Nvidia's own position, stated to CNBC, is that RTX Spark is "roughly equivalent" to an RTX 5070 laptop GPU, not the desktop card . The chip is best understood as a complementary platform — not a 1:1 replacement — for users who want desktop-class RTX graphics and 1-petaflop AI in a portable, power-efficient form factor .

Target Devices and Availability

Over 30 laptops and approximately 10 compact desktop/mini-PC models are confirmed from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE following later . Nvidia's official press release notes that RTX Spark laptops will be as slim as 14 mm and as light as 3 pounds, available in 14- to 16-inch sizes with premium aluminum chassis .

Confirmed early models include :

• ASUS: ProArt P14, P16, and a Mini PC
• Dell: XPS 16 Creator Edition
• HP: OmniBook X 14 and OmniBook Ultra 16
• Lenovo: Yoga Pro 9i
• Microsoft: Surface Laptop Ultra (up to 128 GB RAM, 15-inch mini-LED display, 2,000 nits peak brightness)
• MSI: Prestige N16 Flip AI+

Shipments begin fall 2026 . Nvidia also revealed a multi-generation roadmap for RTX Spark, which PC Gamer reported gives a rare hint of the timeline for next-generation RTX desktop cards, though exact dates remain unconfirmed .

Positioning: Complement, Not Replacement

Nvidia has positioned RTX Spark as a new category — a "PC superchip" that fills the gap between ultraportable laptops (which lack discrete RTX graphics) and bulky gaming notebooks . It is not intended to replace high-end desktop x86 rigs; analysts and press describe it as the first credible challenge to x86 in the mainstream PC market, but acknowledge it will coexist alongside x86 for the foreseeable future .

PCWorld notes that RTX Spark may fundamentally reshape PC building, potentially splitting the market between compact Arm APUs and traditional x86 systems . One Chinese-language analysis observes that while the integrated GPU is RTX 5070-class, the limited memory bandwidth means gaming performance will not match a discrete desktop RTX 5070 in sustained workloads, making the chip most compelling for users who prioritize portability over raw fps .