Nvidia debuted a 64GB configuration of its DGX Spark personal AI supercomputer on October 2, 2026, at approximately $4,999, with the system scheduled to reach store shelves on October 23. The launch gives local AI developers a lower-cost entry point as memory prices rise and supplies remain short, while the original 128GB model now sells for $6,950 after an increase of nearly 75%. Both systems use Nvidia's GB10 Grace Blackwell Superchip, combining a 20-core Arm CPU with a Blackwell GPU and delivering 273 GB/s of memory bandwidth. ConnectX-7 networking allows up to four units to be linked in a cluster. The 64GB version is designed for local inference involving roughly 26 billion to 35 billion parameters. Acer, ASUS, Dell, GIGABYTE, HP, and MSI are launching the configuration, while Lenovo is part of the broader DGX Spark ecosystem but was not explicitly named for this model. Nvidia plans an updated DGX OS release later in October 2026 to simplify cluster setup and inference-model deployment. Larger models may require the 128GB version or a cluster, while further memory-price increases could put pressure on the 64GB model's price and alter the current gap of nearly $2,000.