Short version: we're full. Every certified GPU in the fleet is spoken for, the waitlist keeps growing, and the most common email we get is some variation of "when can we start?" This post is our answer.
What's in the pipeline
We're in conversations with several enterprise IT teams whose refresh cycles land this summer, which means a new tranche of decommissioned hardware is headed for our certification lab over the coming months. As always, nothing gets announced as capacity until it survives the 72-hour burn-in, so we won't quote counts until the benchmarks are logged.
Demand-wise, the pattern is clear: A100s remain the workhorse request, H100 interest is accelerating, and H200 inquiries arrive almost entirely from teams sizing large training runs. If that's you, the earlier we hear about your timeline, the better we can slot you in.
How onboarding will work
New capacity goes to the waitlist in order, with one exception: reserved-capacity commitments (3, 6, or 12 months) get priority within their cohort, since they let us plan acquisitions with confidence. If you're on the waitlist and your needs have changed (more GPUs, a different tier, a harder deadline), send us an update so we're planning around reality.
Thanks for the patience, and for betting on second-life compute. The whole point of this model is that capacity grows as fast as good hardware gets rescued, and there's a lot of good hardware out there.