What Lightning AI sells — and what it buys
Developer seats plus GPU-hours — both first-party (the ex–Voltage Park fleet) and third-party via a Multi-Cloud GPU Marketplace spanning 7+ providers including Lambda, Nebius, Nscale and the hyperscalers. One environment, any cloud, no rework.
Demand routed into other clouds. Lightning is simultaneously a supplier and one of the largest demand aggregators for neoclouds — pre-merger it claimed marketplace arbitrage could cut AI costs by up to 70%.
Scale proof
The bet
That the workflow layer — Studios, orchestration, cloud-agnostic storage — is stickier than any individual GPU fleet, and that owning some supply fixes the marketplace’s chicken-and-egg problem.
Natural counterparty
Enterprise ML teams that want no-DevOps multi-cloud; and on the flip side, capacity providers who want to be listed as a supply source — for a capacity-hungry provider, getting into Lightning’s marketplace is a distribution channel.
Buyer fit
Your team lives in PyTorch and wants portability across providers without rebuilding the stack.
You need a bare-metal, single-provider megadeal at the lowest possible unit cost.
Five fields we track but don’t publish
Everything above is public and verifiable. The fields below change weekly and are verified directly with each provider — they live in a private supply-demand ledger, not on this page.