Skip to content
Source: NVIDIA Blog

🖥️NVIDIA Ships RTX Spark Desktops With 128GB Unified Memory and 1 Petaflop AI

Your laptop can finally stop running your AI agents

TL;DR

NVIDIA's RTX Spark compact desktops from ASUS, Dell, and MSI are shipping with up to 128GB unified memory and 1 petaflop of AI performance. The pitch: a dedicated box runs your local agents around the clock while your laptop travels. Worth a look if you're tired of cloud bills for inference.

NVIDIA's RTX Spark compact desktops are out from ASUS, Dell, and MSI, and the pitch is simple: put your AI agents on a box that never sleeps. Your laptop goes wherever you go, and a dedicated Windows desktop keeps scheduled tasks, agents, and fine-tuning jobs running in the background. For developers, the appeal is the stack. You get native CUDA, Linux tooling through WSL, and agent setups via Hermes Agent or NVIDIA NemoClaw blueprints, plus Unsloth for fine-tuning supported models. Switchyard, now in Hermes, routes tasks between local and cloud models automatically, so you pay for cloud only when you need it. The hardware is a 20-core NVIDIA Grace CPU, a Blackwell RTX GPU with 6,144 CUDA cores, and up to 128GB of unified memory, rated at up to 1 petaflop of AI performance. Ten-gigabit Ethernet helps when you're moving model files and datasets to networked storage. Acer, GIGABYTE, HP, and the Microsoft Surface RTX Spark Dev Box are still coming.

NVIDIA Ships RTX Spark Desktops With 128GB Unified Memory and 1 Petaflop AI — NVIDIA Blog

Key Points

1

The Grace CPU has 20 cores, paired with a Blackwell RTX GPU carrying 6,144 CUDA cores and up to 128GB of unified memory for large local models.

Keep reading — it's free

Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.

Also get
or

Already a member? Sign in

NVIDIARTX Sparklocal AIWindows on ArmAI agents

Comments

Subscribe to join the conversation...

Be the first to comment