Imagine I have only two EC2 virtual machines running the same stateless workload and my priority is very high resilience, followed by cost.
Why would I choose a load balancer to share traffic between the two VMs, each one in a different AZ, instead of using Route53 to distribute traffic across the two VMs, each in a different region? Route 53 significantly cheaper in my estimates.
Assume health checks enabled and TTL is small. Max number of total VMs is 2 (so can’t mix R53 and ELB)
If you own the SSL certificate and those are installed on your EC2 instance or used by your services within EC2, you can give Route53 a try, to split the traffic between your two instances. But it could be, that the load is not well balanced to your instances because of the way, Route53 will send the requests to the EC2 instances. This mechanism is prefered, if you want to have multiple regions available and route the traffic to the nearest region of your client.
Instead of this, a load balancer can, like the name, balance the load within the same region to multiple instances (it doesn't matter of using one or multiple availability zones). If your service should be available also during an update of one EC2 instance or do something like blue/green deployment, you will have a downtime using Route53, until you register your new EC2 instances to Route53. This will do automatically the target group for your...if one instance is failing, your service will be available. Also using ACM within the AWS ELB will abstract you the usage of SSL certificates, because you add them to the ELB and you do not need to distribute them inside your EC2 instances.
Last but not least (and most important): Security
WAF, Login, routing only to specific methods/URIs can handle centrally witin the ELB, not able to do this in Route53.