Quick Facts
- Nvidia’s RTX Spark chip delivers 1 petaflop of AI performance with 20 Arm CPU cores and 6,144 CUDA cores
- RTX Spark laptops from Dell, HP, Lenovo, Microsoft, Asus, and MSI will arrive fall 2026
- The chip combines Blackwell GPU with Grace CPU and supports up to 128GB unified memory
Nvidia announced RTX Spark at Computex 2026, marking the company’s entry into the PC processor market. CEO Jensen Huang unveiled the Windows on Arm platform during his keynote, calling it “the most efficient PC chip ever built.”
The RTX Spark Superchip combines a 20-core Grace CPU with a Blackwell GPU containing 6,144 CUDA cores. The system supports up to 128GB of LPDDR5X RAM with 300 GB/s memory bandwidth and delivers approximately 1 petaflop of AI performance.
“The PC is being reinvented,” Huang said. “For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask — and the PC does the work.”
The integrated GPU matches an RTX 5070’s performance. Users can render 90GB 3D scenes with OptiX and DLSS, edit 12K video, run 120-billion-parameter language models, and play AAA games at 1440p with over 100 frames per second.
Mark Aevermann, Nvidia’s senior director of product management, described RTX Spark as “the most efficient PC chip ever built.” The company developed the Arm-based processor in partnership with MediaTek.
Market Entry Strategy
RTX Spark systems will launch through partnerships with Dell, HP, Lenovo, Microsoft, Asus, and MSI. The first laptops are expected to cost around $2,000 or more, targeting premium, creator, and AI-focused segments.
Huang promised future generations of all Nvidia platforms will include Spark chips, including upcoming Vera Rubin and Rosa Feynman variants with faster memory technologies.
The move positions Nvidia to compete directly with Intel, AMD, Apple, and Qualcomm in the CPU market. Huang noted that Arm-based processors are gaining ground over traditional x86 processors as the CPU market expands into what he projects will be a $200 billion industry.
AI-First Computing
Nvidia positions RTX Spark as enabling AI agents to replace traditional mouse and keyboard interactions. The company envisions users commanding systems through natural language, with AI agents setting goals, calling tools, and refining work using local and cloud models.
The announcement faces challenges from ongoing DRAM and NAND shortages, Windows on Arm limitations, and x86 processor incumbency. However, it represents Nvidia’s expansion beyond its traditional GPU business into the broader PC market.
Read more: Nvidia announces RTX Spark as ‘the most efficient PC chip ever built’