I’d be more excited about these if the pi or hat included >1GbE networking. As is there’s no way to fully exploit that NVMe bandwidth for any networked workload.

Now I have the sudden urge to build a little cluster of these despite not actually having a use case for one