The most expensive mistake on a young project is not the wrong feature, it's the wrong infrastructure. I have watched solo founders pick Kubernetes because they read a blog post and burn three months on operations work they did not need. I have watched teams pick a managed backend because it was easy, and hit a wall the day they wanted anything off-pattern.
My default stack for a new product: Next.js on Vercel, Postgres, and an authentication provider I can replace. That combination scales further than most projects ever go, the deployment story is one command, and the whole system runs locally without ceremony.
I deviate from that default for three reasons, and only three. Real-time collaboration at scale. Heavy compute, machine learning, video, geospatial. And regulatory requirements, where the pragmatic path is purchasing compliance rather than building it.
The five questions I ask every founder to answer before a stack decision: What is the realistic user load in six months? What is the cost ceiling? Who maintains this in two years? What must remain portable if we change our minds? And what does failure look like, how fast can we recover?
Most founders overestimate the first and most engineers ignore the second. Honest answers point at a stack that is smaller, cheaper, and easier to operate than what most teams ship. I default to proven technology until evidence forces me off it.
One more rule: never choose infrastructure during a deadline. Pressure selects for what feels safe today, and that is rarely what is economical tomorrow.