Lightning AI, the Nvidia-backed developer platform, has introduced a multicloud GPU marketplace that lets customers compare, reserve and run graphics processors across both major cloud providers and newer specialist "neoclouds" from a single interface. The service, embedded in Lightning’s existing end-to-end AI development suite, displays pricing and performance data for chips offered by Amazon Web Services, Google Cloud, Voltage Park, Lambda and others. Users can switch providers without rewriting code, selecting spot, on-demand or reserved capacity to optimize cost, performance or geography. Chief Executive Officer William Falcon said the goal is to give AI teams "control, flexibility and speed," while Chief Product Officer Priya Shivakumar added that the marketplace could cut training costs by as much as 70% by reducing unused capacity. The entry tier is free, with small-team subscriptions at $140 per user per month and enterprise pricing negotiated separately.