Nvidia has made a significant stride in the computing landscape with the unveiling of its RTX Spark superchip on May 31st, just ahead of the GTC Taipei event and the Computex 2026 showcase. This innovative chip combines a 20-core Arm-based Grace CPU with a Blackwell RTX GPU, marking Nvidia’s entry into the Windows PC market—a sector where it has historically lacked presence.
The RTX Spark is designed to impress, featuring up to 128GB of unified memory, a claimed 1 petaflop of AI compute capability, and an impressive 6,144 CUDA cores. This powerful chip is set to debut in laptops and compact desktops from major manufacturers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE expected to follow suit. Notably, Microsoft is not merely an operating system supplier in this venture; it has collaborated with Nvidia to develop new Windows security features that will support on-device AI agents alongside Nvidia’s OpenShell runtime.
Previously referred to as N1X, the RTX Spark has been in the industry’s sights for three years. Initial reports surfaced in 2023, indicating Nvidia’s ambition to create Arm CPUs capable of running Windows. The chip has been the subject of speculation throughout 2024 and 2025, with various delays attributed to Microsoft’s gradual advancements in Arm technology and fluctuating notebook demand, ultimately pushing its release into this year.
For eight years, Microsoft’s Windows on Arm initiative was exclusively tied to Qualcomm silicon, a partnership that effectively excluded other chip manufacturers. This exclusivity, which began in 2016, has now come to an end, as confirmed by Arm CEO Rene Haas in January 2024. This shift has opened the door for MediaTek, Nvidia, and AMD to develop Arm-based Windows chips, ready to enter the market once the opportunity arose.
Microsoft’s involvement in the RTX Spark extends beyond its collaboration with Qualcomm on the Copilot+ certification program. The two companies have jointly developed a robust security framework at the operating system level, integrating identity, containment, and policy primitives in Windows with OpenShell’s capabilities. This partnership aims to enhance user privacy and security while utilizing local AI models.
Windows on Arm
During Qualcomm’s eight-year exclusivity, the company showcased the potential of Windows on Arm, yet struggled to achieve significant market traction. In the third quarter of 2024, Snapdragon X laptops sold approximately 720,000 units, representing a mere 0.8% of total PC shipments for that quarter. ABI Research projected that Arm would not surpass 13% of the PC market by 2025, while Qualcomm’s own figures were limited to U.S. retail Windows laptops priced above 0.
A key factor in this underwhelming performance was software compatibility. Microsoft’s Prism technology allowed x86 applications to run on Arm, but many professional tools, such as AutoCAD, were unsupported, and gaming experiences suffered from crashes and rendering issues. While the initial Snapdragon X pitch emphasized battery life as a major advantage, Intel’s Lunar Lake processors soon matched that efficiency, providing users with long-lasting performance on x86 chips that run all Windows applications natively, without the drawbacks of emulation.
The RTX Spark distinguishes itself with its GPU capabilities, CUDA support, and the expansive 128GB unified memory pool. This hardware is tailored for local AI applications, creative professionals, and gamers, rather than prioritizing portability. Adobe is already reengineering its Photoshop and Premiere software for the platform, promising a two-fold increase in AI and editing efficiencies. Over 100 software vendors and game developers, including KRAFTON, NetEase, Remedy Entertainment, Riot Games, and Microsoft Xbox, have pledged their support for this new platform.
However, two persistent challenges that have plagued Windows on Arm for nearly a decade remain unresolved. The first is the reliance on x86 emulation, which continues to hinder performance for applications lacking a native Arm build. While Nvidia’s CUDA and RTX stack operates natively, the legacy software and peripheral drivers still present a hurdle. The second challenge lies with Microsoft itself; delays in the development of the next-generation Windows on Arm platform were cited as a primary reason for the N1X setbacks. Developers will not receive a comprehensive overview of the Windows agent features until Microsoft’s Build keynote on June 2nd and 3rd, just days after the chip’s announcement.
As for pricing, details remain scarce. The only reference point is the DGX Spark, which starts at ,999 but has recently approached ,000 due to supply chain pressures. Given the costs associated with LPDDR5X memory and TSMC’s 3nm manufacturing, it is likely that the RTX Spark will be positioned at a premium price point, far exceeding the sub-0 target that Qualcomm aimed for to expand Arm’s market reach.
With the introduction of the RTX Spark, Nvidia is stepping into a realm that Qualcomm could only attempt to penetrate, armed with a unique asset and facing the compatibility and operating system challenges that no amount of computational power can resolve independently.
