There is a strong pull toward interesting technology, particularly on greenfield projects. It is usually the wrong instinct.
The novelty budget
Every unfamiliar technology in your stack costs you the same things: hiring difficulty, debugging time, unknown failure modes, and a smaller pool of answers when something breaks at 3am.
You can afford perhaps one or two of these per system. Spend them on the part that actually differentiates your product.
What this means in practice
For most business systems, PostgreSQL or MySQL, a mainstream server-side framework and a boring deployment target will outperform anything more adventurous — not in benchmarks, but in the metric that matters: how long the system keeps working after the people who built it have moved on.