I grew up with blackouts, now I put GPU clusters where power exists

I grew up in LATAM, scheduling life around blackouts while the region sat on abundant energy. Now I live in San francisco, where the AI economy's bottleneck is power, so I'm building ATI. We deploy standardized GPU nodes into industrial buildings that already have electrical service. We skip the interconnection queue and aim for standard 2TB of GPU memory (infiniband pair for 4TB), 99.5% uptime SLA, 3/6/12 month terms and priced under big cloud's reserved rates. We want to work with smaller teams to use our deployments and understand their preferences so we can adapt. We are deploying US first but also accessing hydro power in the south american region like the Itaipu Damn in Paraguay and other damns in Ecuador. The ask is founding teams and attempt to lock price for 12 months with a direct line to me and a real vote on config (H200 vs MI325x for memory bound work). Or any other config. If you'd reserve a node, or almost would and can also tell me why not I would greatly appreciate it: gmiguel@antifragil.ai. I'll be in the comments.