Nvidia RTX Spark laptops are now on sale, bringing local AI compute of up to 1 petaflop and up to 128GB of unified memory to slim Windows PCs. Systems from ASUS, Dell, HP, Lenovo, Microsoft and MSI start at $2,399, with additional models expected later. For companies, this matters because workstation-class AI moves from shared servers to an employee laptop.
Superchip specs and six launch events
The launch covers premium portable PCs built around a single superchip that combines an Nvidia Blackwell RTX GPU with an Nvidia Grace CPU. Nvidia positions the platform for local AI work, content creation and gaming in one device, supported by the full Nvidia AI platform. The company also points developers to tested recipes, playbooks and resources for running local AI on RTX Spark.
The hardware formula centers on unified memory of up to 128GB, shared between processor and graphics. That design removes the separate video-memory ceiling typical for laptops and lets large models and creative projects stay resident locally. Combined throughput reaches up to 1 petaflop of AI compute, which explains the focus on inference, fine-tuning experiments and media workflows without constant cloud round trips.
Nvidia marked availability with six retail launch celebrations worldwide, including Berlin, Beijing, Istanbul, Seoul, San Francisco and Taipei. In Berlin, the event took place at XPERION inside Saturn at Alexanderplatz during the Festival of Lights, with RTX Spark-themed lighting on the Berlin TV Tower. Founder and CEO Jensen Huang made a surprise appearance there, greeted visitors, distributed popcorn and gave away five personally signed laptops.
What local AI laptops change for business
For small teams and independent specialists, RTX Spark lowers the barrier to working with large local models, video and 3D tasks on a single machine. A starting price of $2,399 keeps it in the premium segment, yet it replaces a combination of a thin laptop plus rented cloud GPU time for many routine jobs. Large organizations gain a standardized Windows option for AI developers, designers and engineers who need portability with predictable performance.
The limits deserve attention before purchase: actual memory size, performance and battery life will vary by vendor and configuration, and the source does not disclose detailed specifications per model. Buyers should verify supported frameworks, driver maturity and corporate management tools for Windows fleets. This launch alone does not prove that local execution will be cheaper than cloud infrastructure for sustained heavy training loads.
A practical marker to watch is the release of additional systems beyond the first six vendors and independent tests of local model performance on 128GB configurations. Stable availability in retail channels and published workflows from Nvidia will show whether RTX Spark becomes a standard tool for AI teams. If corporate orders follow the fan events, portable local AI will shift from demonstration to procurement plans.
