Perhaps this changes the calculus. If you don't need as much of an ops team to go private cloud. Then you might not need to be as large before it makes sense.
By a private cloud making sense, I mean the lower bound for being an Oxide customer, i.e. having the budget to buy the infrastructure with enough headroom for usage spikes and expected near-term growth. The reasons why you otherwise would go public cloud when you’re still small. The reduced need for an ops team with Oxide was already factored in.
So uber did private cloud, then went to public cloud recently because the balance was not worth it ( and they probably had enough scale to negotiate a good price, like netflix ) . If they were on something like oxide which reduced infra management costs a lot staffing wise and negotiated a bulk deal, would’ve they gone the public cloud route?