* arm instances for Lambda workloads. Do you know if you can choose your architecture when deploying a function? I'd be curious to see the performance of node on x86 vs aarch64
It's the price for a not-very-specific vcpu. Your performance could be better on the A family, more than offsetting the price difference. Of you're really considering the choice - you should test your workload on both. Or maybe they save you time in a different way - for example you save time not having to run a virtualised linker after cross-compiling.
This is really cool we'll likely see custom arm servers running in AWS as a way to reduce cost / improve scalability.