I’m planning to order a DGX Spark from the NVIDIA Marketplace. I’ll be staying at a hotel in Los Angeles during my trip to the U.S., and I’d like to have the unit delivered to my hotel while I’m there.
Before I place the order, I wanted to ask a few questions:
Is there anything specific I should be aware of when placing an order to a hotel address in the U.S.?
Are there any important details or verification steps required for international buyers?
For those who have purchased DGX Spark before — how long did delivery take after placing the order?
Have there been any stock or availability issues recently?
I just want to make sure everything goes smoothly and that I can receive the unit during my stay.
Any advice or experience would be greatly appreciated!
Unless things have COMPLETELY changed, I wouldn’t risk that! This thing was supposed to have come out in March and lots of people were waiting for months. The only brick and mortar store that sells them in the whole country is Microcenter, so if you’re not in driving distance (sorry), I don’t think there’s any way to guarantee yourself one at any given time from what I see.
How has the system performed in your use cases — especially in terms of CUDA and PyTorch support? Have you tried running any robotics or simulation workloads such as Isaac Sim, Omniverse extensions, or reinforcement-learning environments on it?
I’m also curious about its performance for large-scale model training and inference — does the unified memory and FP4 support make a noticeable difference in real-world tasks?
Any insights on stability, thermal behavior, and compatibility with NVIDIA’s SDKs (like TensorRT, Isaac, or JetPack tools) would be really helpful.
I’ve been extremely impressed with everything, and I did a lot of research and was expecting this to be much slower at inference than it is. Mine also stays cool to the touch and silent. I’m getting 40 tps on gpt-oss:120b, a very impressive model. Qwen3-coder:30b is giving me 80 tps!
I’ve run a few playbooks and a couple fine tunes, so many things to try, but this thing really shines at running LLMs that are made for it (and the new gpt ones are kinda big deals).
Check the Microcenter next to your hotel and reserve one then pick it up in person.
Alternatively if you’re on a business trip you may ask to use your company address or a colleague/ friend’s home address you trust.
Also, I reserved my DGX Spark Bundle in Sept, and received it early this week to my home address, just finished the updates in between meetings and planning on finishing my personal setup this weekend.