But the way their offerings are structured means it will be quite expensive to run at scale without a multi cloud setup. You can't globally cache the results of a worker function in CDN, so any call to a semi dynamic endpoint incurs one paid invocation, and there's no mechanism to bypass this via CDN caching because the workers live in front of the CDN, not behind it.
Despite their media towards lowering cloud costs, they have explicitly designed their products to contain people in a cost structure similar to but different than via egress fees. And in fact it's quite easily bypassed by using a non Cloudflare CDN in front of Cloudflare serverless.
Anyway, I reached a similar conclusion that for my app a single large server instance works best. And actually I can fit my whole dataset in RAM, so disk/JSON storage and load on startup is even simpler than trying to use multiple systems and databases.
Further, can run this on a laptop for effectively free, and cache everything via CDN, rather than pay ~$100/month for a cloud instance.
When you're small, development time is going to be your biggest constraint, and I highly advocate all new projects start with a monolithic approach, though with a structure that's conducive to decoupling later.