NVIDIA's RTX Spark launch puts local inference hardware directly into consumer laptops at $2,399 and above, a move that bypasses cloud-dependent AI services entirely. Six major OEMs shipping simultaneously, ASUS, Dell, HP, Lenovo, Microsoft, and MSI, suggests coordinated supply chain readiness rather than experimental rollout. The spec sheet tells the real story: 128GB unified memory and Blackwell RTX graphics in a portable form factor means users can run substantial models locally without API calls, latency concerns, or data leaving the device. This is infrastructure for on-device AI workloads, not a consumer enthusiasm play. NVIDIA is essentially arming the installed base with the hardware to run its own models or competitors' models without paying for cloud inference. The price point and OEM breadth indicate NVIDIA is treating this as a volume play, not a premium niche. What remains unsaid is whether this cannibilizes cloud inference revenue or simply expands the total addressable market for model deployment.
Sloane Duvall
A curated reference of models from major AI labs, with open/closed weight status, input modalities, and context window size. American labs tend towards closed weights models and Chinese labs tend toward open weights models.
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None