On the S3 side, you need to factor in Intelligent Tiering. We're not paying S3 Standard prices for this--recall that this is a mix of mostly cold data. It's a 10x difference per GB on average between EBS and our mix of S3 Intelligent Tiering.
Add in that your RDS instance needs to be high availability like S3 is (and like our RDBMS is). That means a multi-AZ deployment. Multiply your RDS cost by two, including the cost of storage. That still isn't as good as S3 (double price gets you a passive failover partner in RDS PostgreSQL; S3 is three active-active AZs), but it's the best you can do with RDS. We're north of $500/month now for your example.
Add in the cost of backups, because your RDS database's EBS volume doesn't have S3's durability. For durability you need to store a copy of your data in S3 anyway.
Add in that you can access S3 without cross-AZ data transfer fees, but your RDS instance has to live in an AZ. $0.02/GB both ways.
Add in the personnel cost when your RDS volume runs out of disk space because you weren't monitoring it. S3 never runs out of space and never requires maintenance.
$500/month for 1TB of cold data? We were never going to pay that. I won't disclose the size of the data in reality but it's a bunch bigger than 1TB. We host an on-prem database cluster for the majority of things that need a RDBMS, specifically because of how expensive RDS is. Things probably look different for a startup with no data yet, blowing free AWS credits to bootstrap quickly, but we are a mature data-heavy company paying our own AWS bills.
As a final summary to this rant, AWS bills are death by a thousand papercuts, and cost optimization is often a matter of removing the papercuts one by one. I'm the guy that looks at Cost Explorer at our company. One $500/month charge doesn't necessarily break the bank but if you take that approach with everything, your AWS bill could crush you.
Add in that your RDS instance needs to be high availability like S3 is (and like our RDBMS is). That means a multi-AZ deployment. Multiply your RDS cost by two, including the cost of storage. That still isn't as good as S3 (double price gets you a passive failover partner in RDS PostgreSQL; S3 is three active-active AZs), but it's the best you can do with RDS. We're north of $500/month now for your example.
Add in the cost of backups, because your RDS database's EBS volume doesn't have S3's durability. For durability you need to store a copy of your data in S3 anyway.
Add in that you can access S3 without cross-AZ data transfer fees, but your RDS instance has to live in an AZ. $0.02/GB both ways.
Add in the personnel cost when your RDS volume runs out of disk space because you weren't monitoring it. S3 never runs out of space and never requires maintenance.
$500/month for 1TB of cold data? We were never going to pay that. I won't disclose the size of the data in reality but it's a bunch bigger than 1TB. We host an on-prem database cluster for the majority of things that need a RDBMS, specifically because of how expensive RDS is. Things probably look different for a startup with no data yet, blowing free AWS credits to bootstrap quickly, but we are a mature data-heavy company paying our own AWS bills.
As a final summary to this rant, AWS bills are death by a thousand papercuts, and cost optimization is often a matter of removing the papercuts one by one. I'm the guy that looks at Cost Explorer at our company. One $500/month charge doesn't necessarily break the bank but if you take that approach with everything, your AWS bill could crush you.