Luke's stats show ~14% but they lag by a month. They cut out duplicates where possible, and treat all AWS as one node. I don't think it's possible to be more pessimistic than 14%. The 25% we saw yesterday appeared to be Umbrel users updating and getting counted twice, but that isn't happening today with the Start9 update so that theory is just wrong afaict unless I have some misunderstanding about how Umbrel nodes differ from Start9 nodes.