I've been working on a sepic converter. There are some great parts out there in tiny packages.
Then, for example, I find the fets need a square inch of 2oz copper (to achieve 78C/W), to get the heat out. And the diodes, and the sense resistors, and ... A square inch, I can fit another 4 inductors in that space.
What is the point of making parts that small? I'm looking at a TO220 fet now because it needs less pcb space than a superSOT-6 device.
It's because almost all commercial boards are machine assembled these days, so small size is not only not a problem, it's a major benefit in apps were every square mm counts, and that's more common than you'd think. The majority of the market want small parts, so that's the direction the market heads.
There's often a "minimum price" point that in any given year for a cost-sensitive design will dictate using the cheapest parts.
At the same time there's a "minimum size" selling point that drives products to use smaller parts than their competitors because despite the slightly higher price of the smaller parts, the smaller product will sell better.
Then there's the other "maximum feature" selling point that requires products to have the maximal number of features to differentiate them from the rest of the pack so they can charge a slight premium over the cheapest ones. More features usually means more parts. (At a very minimum, more buttons... witness the ungodly 113-button remotes.)
These feed back to the technologies to give a gradual but overwhelming move towards smaller and smaller parts and higher and higher power densities. Power density may not be an actual goal but it is a result of all the other trends, and from a part or PCB layout perspective it can seem completely bizarre as you point out. You'd think all those consumers clammering for smaller stuff would also want lower power demands for stuff... but the heavy drive towards high power density batteries has in fact caused some concern because at the rate we're going, we're going to run out of some of the metals used in batteries in decades, not centuries or millenia.
In fact the fact that you're making a SEPIC converter for what used to be a tradtional inverter/transformer application means that you're going towards some of the same goals as the mass market.