Nvidia announced at IFA that its long-awaited RTX Spark N1X system-on-chip will ship in two configurations when it launches in October. The top-end version pairs a 20-core "Grace" Arm CPU with a Blackwell-based GPU containing 6,144 CUDA cores and supports between 24GB and 128GB of unified LPDDR5X memory; it is aimed at both laptops and mini PCs. A second, slightly cut-down version uses an 18-core Grace CPU with a 5,120-core Blackwell GPU and 24GB to 32GB of memory, and will be laptop-exclusive at launch. Sources differ on the naming implications: TechPowerUp reports Nvidia confirmed both chips are simply "N1X" configurations with no separate "N1" chip, while Tom's Hardware suggests the "N1X" branding hints a lower-end "N1" chip could arrive later.
OEM partners confirmed for launch include Acer, Asus, Dell, HP, MSI, Lenovo, Microsoft, and Gigabyte, according to Wccftech. Specific systems unveiled at IFA include Lenovo's Yoga 9n 2-in-1 convertible, described as just 14mm thin, and Acer's SFF RTX Spark mini desktop. Microsoft's Surface Laptop Ultra, previewed earlier at Computex, is also expected among the RTX Spark lineup. Laptops using the chip will run Windows 11 for Arm and be classified as Copilot+ AI PCs. Nvidia says availability will vary by OEM and region, with some launching in limited markets first before going global.
Nvidia is promoting bold performance claims for RTX Spark, including 100 frames per second at 1440p in gaming with ray tracing and DLSS enabled, supported by DLSS 5, DLSS 4.5 Ray Reconstruction, and multi-frame generation up to 6x. PCWorld cautions that these figures rely heavily on frame generation technology rather than conventional rendering, urging skepticism until independent testing occurs. For content creation, Nvidia says the platform can render 90GB 3D scenes using OptiX and ray tracing, with hardware-accelerated 4:2:2 video support. On the AI side, Nvidia touts support for agentic AI models up to 120 billion parameters with 1 million-token context windows, tied to Windows' new Agent Framework. The company also claims major game publishers—including EA, Ubisoft, Konami, Capcom, CD Projekt Red, and Xbox—are on board to support the platform.
Pricing has not been disclosed by Nvidia. Tom's Guide offers estimated predictions based on components and market comparisons: the baseline configuration (18-core CPU, 5,120 CUDA cores, 24GB memory) could start around $1,999 to $2,499, while a maxed-out 128GB configuration could run $4,000 to $5,000 or more. PCWorld similarly expects systems to cost several thousand dollars given the memory configurations involved, though exact figures from OEMs remain unconfirmed.
The RTX Spark platform uses a unified memory architecture similar to Apple's M-series chips, including the recently announced M6 and M5 Ultra, and AMD's Strix Halo and upcoming Gorgon Halo designs, according to Tom's Hardware. Microsoft is reportedly preparing a Windows 11 feature called IntelligentCarveout, per TechPowerUp, that would let systems manage unified memory pools across CPU and GPU workloads in chips like RTX Spark and AMD's Ryzen Halo lineup, including configurable partitions of that shared memory. The N1X launch also ends Qualcomm's exclusivity as the sole Arm chip vendor for Windows PCs, introducing a major new silicon competitor to the Windows on Arm ecosystem.
Not all coverage is uniformly optimistic. PCWorld notes that Windows-on-Arm laptops using Qualcomm's Snapdragon chips had a tepid reception since their 2024 debut, with early gaming performance shortfalls offset somewhat by Intel and AMD's efficiency gains since then. The outlet also raised questions about Nvidia's broader commitment to consumer products given its data center business. Tom's Guide, by contrast, frames RTX Spark as a potential turning point that could give Apple's MacBook Pro lineup its first serious competitor, combining productivity performance with legitimate PC gaming capability and claimed all-day battery life—though it stresses this outlook depends on Nvidia's performance claims holding up in practice.