Does the competition between L1s disprove the Fat Protocol thesis?

For those not initiated, the Fat Protocol thesis was posited by Joel Monegro in 2016. At a high level, it posits that while in previous iterations of the internet, value accrued to the application layer (google, Amazon, Fb) rather than protocol layer (http, smtp) this will be reversed. Full post here: https://www.usv.com/writing/2016/08/fat-protocols/

This theory seems vindicated in 2022. Layer 1 blockchain are more valuable with than Dapps, with 4 smart contract platforms in the top 10 largest cryptocurrencies by market caps. However, crypto markets are notoriously inefficient and volatile, and I wonder if this is an uneasy equilibrium. Comparisons to the internet bubble are a tired heuristic, but the telecom bubble followed similar logic to the layer 1 build up. “If we own the infrastructure, we don’t need to pick the specific application winners”

My concern is that we haven’t seen layer 1s have enduring “moats” or competitive advantages to give pricing power. If you can save 90% on fees, customers don’t seem to care if their dex is on ethereum or avalanche. Due to these market dynamics, gas fees are likely to asymptotically approach zero across the ecosystem, therefore reducing protocol value accrual. Similarly, as a tokens value increases, the networks utility decreases since fees are denominated in the native token. For instance, if ethereum increases from $3k, to $30k, gas fees will also increase 10x and there will likely be an exodus of DApps to other ecosystems.

Has anyone thought this through before from first principles? Would love to read a blogpost or podcast addressing these concerns. As it stands, I am concerned that value will accrue to applications and great sums of money will be lost on investments in the protocol layers.