Even looking at comparative offerings from different providers, they're really lackluster [1], and in some cases, even more expensive [2].
What could be contributing to those large numbers?
[1] - Memorystore only allows vertical scaling, no support for GCP managed Redis clusters, no hybrid functionality (only instances in the same VPC network can access Memorystore, whereas ElastiCache offers all of this.
To be fair, it seems Memorystore instance is not the same gb-for-gb as Elasticache, since Memorystore gives you extra memory overhead, and on aws you have to manage it yourself. Memorystore also has free network egress/ingress. There is also a RedisLabs offering on Google Cloud with integrated billing.
if you run Redis you need at least 25% of extra RAM on top the instance memorysize if you want to avoid a lot of nasty OOM scenarios. Memorystore gives this memory by default and in aws you need to tweak the reserved-memory. https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/...
Disclaimer: I work at Google Cloud.