The San Francisco Compute Company (SF Compute) runs large-scale, vetted GPU clusters and operates a compute marketplace where buyers reserve H100/H200 capacity by the hour and sellers resell unused compute back into the market with no long-term lock-in. - View it on GitHub