serverless scaling

Can Lambda MicroVMs work as self-hosted GitHub Actions runners?

I'm looking at Lambda MicroVMs for runners that start when a job arrives and scale to zero afterward. Is that practical, and what tradeoffs should I compare with…