A high-memory AI workstation that fits in a laptop
RTX Spark Windows PCs began arriving October 17, with systems from ASUS, Dell, HP, Lenovo, Microsoft and MSI starting at $2,399. NVIDIA lists up to 128GB of unified memory and up to one petaflop of AI compute. The hardware pairs a Blackwell RTX GPU with a Grace CPU, bringing the company’s accelerated-computing platform into premium portable machines.
For local model users, the memory figure is the key promise. Large models and multimodal workloads can be constrained by how much memory is available to the GPU; a unified pool may make some experiments possible without sending every prompt to a cloud service. It does not mean every 128GB model will run quickly, nor does a peak compute figure predict end-to-end performance for a particular model.
The buying question is workload, not the launch spectacle
The products are positioned for AI development, creative work and gaming, but the launch post does not provide a comparable benchmark suite across vendors. Buyers should compare the configuration they can actually purchase, sustained performance, cooling, battery life, memory allocation and support for the model stack they use.
At a starting price above $2,000, this is a specialized local-compute option rather than a general laptop upgrade. Its appeal will depend on whether having a large memory pool on the desk—or in a bag—saves enough cloud expense and iteration time to justify the premium. More systems are expected to arrive from the announced PC partners.
Source published 2026-10-17. Coverage is based on the maker’s announcement and demonstration.