Sometimes I like to think of technology in terms of roads and buildings.
Roads start quite simply, just a dirt track, with minimum effort. Buildings set up around roads and take more time to set up.
The difference? Roads get locked in, you can upgrade and improve a road, but a seldom moved or widened. Houses seem hard to change, but in reality, they can be knocked down and rebuilt with little to no impact.
Too often, tech debt is the result of roads; things that didn’t seem significant at the time but provide the fundamentals for other decisions. I'm not specifically talking about things like AWS vs Azure, but talking even more fundamental (e.g., programming language, first X people hired). These decisions can be taken for granted, and seem like they can be changed, but actually, are harder than you think.