While the Fat Protocol theses clearly holds true when comparing Ethereum with the combined market cap of all the projects built on top of it, it’s not as straight forward on the layers above. Projects on higher layers hold more market cap on aggregate, but projects on lower layers have the higher average market cap.
To some degree, the lower total market cap might well be due to layer 2 & 3 technologies still being in development themselves, and therefore not being used by high-layer projects yet. But not all layer 4 protocols will need all layer 2 & 3 solutions to work. From looking at average market cap, the premise of the Fat Protocol thesis still seems to hold though.
Binance Coin on the other hand shows that significant value accrual can take place at the application layer as well, outperforming all individual layer 2 & 3 coins. Even in that case, it’s not outperforming the base layer protocol though.
Then still, the cryptocurrency space is still young, and value is based more on speculation than fundamentals, so we’ll probably have to revisit this in year or two.