31/07/2026
Most founders can't tell you where their data actually lives. They know a product name. That's it.
Here's why that matters more than it sounds like it should.
Storage and compute used to be welded together. Pick a database, you picked how you'd query it, forever. Object storage (S3, or self-hosted alternatives like MinIO) broke that link. Your files sit on their own, in an open format — and any engine can read them.
We built a lakehouse on exactly that split. Hundreds of tables, five layers, storage nobody thinks twice about. Boring, on purpose.
Here's the trap: a managed cloud warehouse is fast on day one. Every table you write against its proprietary format is a small rope tying you to that vendor, though. Cut it later and you're not moving files. You're rebuilding.
Fintechs feel this hardest. Ten thousand transactions a day can become two million inside one funding cycle. If your storage choice made switching engines expensive, that's exactly the week you find out.
Open storage doesn't make switching free. It makes it possible.
Not every startup needs this on day zero. A ten-person team on a managed warehouse isn't wrong. The mistake is not knowing you made a choice — until undoing it costs more than building it did.
Next up: who's actually allowed to change your tables. Governance starts one layer up from storage.
https://yacooti.substack.com/p/where-your-data-actually-lives-and?r=2aiw1l
