It just a time limit of the life of a single MicroVM.
Using this for a long lived "developer environment" would be extraordinarily expensive anyhow. Scaling the vCPU + RAM cost of these to the same shape compute optimized Graviton On-Demand EC2 instance (16 vCPU x 32 GB RAM) shows about 4x the cost.
So don't do that. Just use an EC2 instance.
But these have near instant suspended/resume, and they even have vertical scaling of the ram, which is a great feature that’s not very common.