NVIDIA DGX Cloud Lepton alternative: a neutral GPU marketplace

NVIDIA DGX Cloud Lepton is NVIDIA's marketplace for GPU compute, launched in May 2025. If you are looking at it, you probably want what most GPU buyers want: one place to find and buy capacity instead of juggling separate accounts. This post covers what Lepton does well, where its scope ends, and why Opengrid is building a provider-neutral alternative.

Up front: Opengrid is pre-launch. The market view ships first and the routing API is in development. Where this post describes Opengrid, it describes what we are building, not something you can sign up for and use today.

What DGX Cloud Lepton does

Lepton is a marketplace across NVIDIA's cloud partners. Instead of going to each partner separately, you find and buy capacity from them through one place, backed by NVIDIA.

That is a real convenience. Every new provider normally means a new account and contract, a new API and tooling, a new bill to reconcile, and fresh testing of who is actually reliable. A marketplace that puts several providers behind one front door removes a lot of that work, and having NVIDIA behind it gives buyers some confidence in the partners listed.

Where Lepton's scope ends

Lepton's design follows from who runs it. It is a marketplace built by the chip vendor, for the chip vendor's partners. Three things follow from that:

• NVIDIA partners only. Lepton covers NVIDIA's cloud partners. SemiAnalysis tracks 209 GPU clouds, and a partner marketplace is, by definition, a subset of that market.
• Not neutral. The company running the marketplace also sells the chips. That isn't a criticism. It is simply the structure, and it means Lepton is not built to be an independent view across every cloud and every chip.
• Partners set prices, and there is no benchmark. Prices on Lepton are set by the listing partners. The marketplace doesn't tell you whether a price is fair compared with the rest of the market. That matters when the same H100 lists anywhere from about $1.25 to $11.07 an hour depending on who you rent from.

Lepton also doesn't route by workload. You still choose what to buy rather than describing a job and having it placed.

How Opengrid is different

Opengrid is built to be neutral by design. It doesn't own GPUs and doesn't make them, so it has no reason to favor one provider or chip. The goal is the whole market: every GPU cloud and every chip, side by side.

Opengrid is built in three layers, in order:

1. See the market (now): pricing, availability, capacity and regions across providers in a single live view.
2. Route the workload (next): one API key, every GPU cloud behind it. Opengrid evaluates providers on price, availability, location and reliability, then places the job.
3. Trade compute (later): real routed jobs produce real prices, a live price for every GPU and region.

With routing, you describe the job instead of picking a listing:

POST /v1/route · 8× H100 · US · < $2.50/hr · done within 12 hours

Illustrative request. The routing API is in development.

Because the market view comes first, every placement can be checked against the full spread of prices, not just the prices on one marketplace.

DGX Cloud Lepton vs Opengrid at a glance

	DGX Cloud Lepton	Opengrid
Status	Live, launched May 2025	Pre-launch. Market view first, routing API in development
Run by	NVIDIA	An independent company, not a GPU vendor or cloud
Coverage	NVIDIA's cloud partners	The goal is every GPU cloud
Provider neutral	No	Yes
Buy in one place	Yes	Yes
Fair-price benchmark	No	Yes, from the live market view
Routes by workload	No	Yes, that is the core product

Which should you use?

Use DGX Cloud Lepton today if NVIDIA's partner network covers what you need and you want a live marketplace backed by NVIDIA right now.

Join the Opengrid waitlist if you want a neutral view across every cloud and every chip, a benchmark for whether a price is fair, and routing that places the job for you. GPU prices now move by double digits in weeks, and capacity comes and goes, so a wider view of supply is worth more when the market is tight, not less.

To see where Lepton sits among the other options, read our comparison of GPU cloud aggregators and marketplaces, or compare it with Shadeform, another multi-cloud option.