Nvidia’s RTX Spark seen as direct challenge to Apple in local AI

Nvidia just drew a line in the sand against the last company most people expected: Apple. With the launch of its RTX Spark Arm-based superchip at Computex Taipei, the GPU titan is moving beyond its data center fortress and into the personal AI computer market, a space where Apple’s M-series chips have been sitting comfortably for years. The RTX Spark is designed to run 120-billion-parameter large language models locally, with a context window stretching up to 1 million tokens. For perspective, that’s enough to process roughly the length of several novels in a single conversation, all without sending a byte to the cloud. Nvidia’s consumer AI play The RTX Spark represents Nvidia’s bet that the AI workloads currently confined to massive server farms will increasingly migrate to personal machines. The chip targets high-end laptops in the $3,000 to $4,000 range, aimed squarely at developers, researchers, and power users who want to run sophisticated AI agents without a cloud dependency. The chip supports unified-memory inference in laptops, a design philosophy that Apple pioneered with its M-series architecture and one that eliminates the bottleneck of shuttling data between separate me...