I'd like to see if we can use VMs only when Digital Ocean has spare resources available (CPU). Google Cloud charges half the normal rate, but perhaps we could bid on VM that would run until those resources are needed or until 24 hours have run out.
For example, an optimized droplet for CPU will cost $20/mo.
If I opt for preemptible VMs, perhaps you can charge $0.50/day for the VM use but only if resources are available. If resources are not available, you suspend the VM. Att that point, just charge us for storage.
Seems like a good way to fully utilize your data center resources while giving cost conscious customers a way to save money for parallel jobs such as 3D rendering, data crunching, distributed computing (SETI@Home), etc.