That number is based on experience with Grist running with the CPU and memory resources we give it in our hosted service. If you have far more records, Grist as it is today won’t be a super experience (we have ambitions to iterate on that of course). The main architectural constraint currently is the spreadsheet side of Grist (tracking dependencies between cells etc).
When self-hosting, you can scale CPU and memory, so if things start getting sluggish you have options.
2 Likes