Nvidia presents RTX Spark superchip - Windows to become an AI platform
Nvidia's RTX Spark superchip forms the basis for a new era of AI computing - at least according to the manufacturer. The Windows-on-Arm platform combines an Arm processor, a Blackwell GPU and up to 128 GB of RAM.
AI agents could revolutionise the interaction between humans and computers. Instead of a mouse and keyboard, natural language and autonomous software agents will dominate in the future - this is how Nvidia CEO Jensen Huang outlined his vision at Computex 2026. To underpin this idea technically, Nvidia presented the RTX Spark superchip https://www.nvidia.com/en-us/products/rtx-spark/.
Familiar chip, new stage
The chip combines 20 Arm CPU cores from Mediatek, a Blackwell GPU with 6144 CUDA cores and up to 128 GB of LPDDR5X memory with a bandwidth of 300 GB/s. It is therefore a system-on-a-chip (SoC). The CPU and GPU communicate via NVLink C2C. According to Nvidia, the performance is sufficient for language models with 120 billion parameters and context lengths of up to one million tokens - with a TDP of around 80 watts. The chip is based on the GB10, which is already used in the DGX Spark.
Over 30 laptops and ten desktops at launch
The platform is initially aimed at laptops. Partners such as Dell, HP, Lenovo, Microsoft, Asus and MSI are planning corresponding devices. Nvidia promises tandem OLED displays with G-Sync, all-day battery life and aluminium housings with large glass touchpads. The above specifications apply to the full version. There will be several versions of the SoC. Performance is said to be just as high on battery power as on mains power - a feature familiar from Windows-on-Arm and Apple Silicon devices. Compact desktop PCs should also benefit. Nvidia is expecting over 30 laptops and around ten desktop systems at launch.
Gaming, creative work and Adobe integration
In addition to AI applications, Nvidia is positioning the RTX Spark as a platform for gaming and creative work. The company promises 100 FPS in 1440p, supported by DLss 4.5 and Multi Frame Generation. The large RAM should be able to handle 3D projects and high-resolution video material such as 12K 4:2:2 content with ease. Adobe is developing Photoshop and Premiere for RTX Spark, and both programmes will be controlled by AI agents via the Model Context Protocol. Software manufacturers such as Blender, DaVinci Resolve and Cinema4D are also planning native support.
Source: Youtube / Nvidia
Windows becomes an agent platform
Nvidia is working with Microsoft on an OpenShell framework that extends Windows with security mechanisms for AI agents. These are intended to ensure that local models only access data and tools for which the user explicitly grants permission. Microsoft plans to announce further details at a conference.
Many promises, few figures
Nvidia has not yet presented any concrete performance comparisons with Intel, AMD, Apple or Qualcomm. There are also no details on the Prism emulator, which is designed to run older x86 software on the Arm platform. Its efficiency is crucial for the performance of existing apps and games. There is also no information on battery life in various scenarios and prices.
Assessment
Nvidia dominates the AI market not only with powerful chips, but also with CUDA, long-standing developer relationships and the ability to create and dominate new market segments. The company now wants to transfer this recipe for success to the Windows PC. RTX Spark is less a chip than a platform.
Whether this succeeds depends on factors that Nvidia has not yet revealed: prices, suitability for everyday use and, above all, whether the OpenShell framework also works on hardware from other manufacturers. If it remains tied to Nvidia silicon, this could slow down its spread.
One key question remains for users: Will a PC that performs tasks independently in the background really change everyday working life - or will this vision remain a theory for the time being? The answer will only be provided by the devices that should be available from autumn 2026.
From big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.
From the latest iPhone to the return of 80s fashion. The editorial team will help you make sense of it all.
Show all