Is it surprising that relatively niche configurations (especially ones which clearly need specific hardware per compute) are finite in aws? But imagine aws runn

Is it surprising that relatively niche configurations (especially ones which clearly need specific hardware per compute) are finite in aws? But imagine aws running out of m4.X-Large or something. That would be interesting. What scale would you need to run out of the most common instance in a top availability zone? 10k? 100k?