Most probably liquidity is the biggest challenge. Big enterprises have to bank the channels for liquidation of balances. So, you probably end up with several centralized kind of banks. Not sure it’s entirely bad though.

Also another concern (besides my other concerns) is the dust limit.

An interesting but not well-known fact of the Lightning Network is that when you make a very small payment (let's say, 100 satoshis), the payment isn't actually "trustless". This is because an HTLC of 100 satoshis can't be put on-chain due to the dust limit, as it's assumed it wouldn't be economical to spend such outputs and thus are considered spam. Thus if your partner tries to cheat you out of these funds you won't be able to settle it on-chain via force-close.

Right now this isn't an issue, the dust limit is a few hundred satoshi and you wouldn't cry over losing such funds (and you assume your channel partner wouldn't try to steal such a small amount from you). But the dust limit will increase as base layer usage increases, I don't know to what extent but what if some day the dust limit is $50? Then the vast majority of LN payments would not be trustless.

Bitcoin and lightning won't scale to have every person on the network to have their own lightning node in today's form. Various soft forks could improve this but that's extremely hard. Likely many people willing only ever be Wallet of Satoshi users and not lightning users.

It'll be difficult to onboard billions of people without some sort of channel factory, either through eltoo, or more interestingly a recent proposal by John Law to add hierarchical channels without a soft-fork (https://github.com/JohnLaw2/ln-hierarchical-channels/blob/main/hierarchicalchannels12.pdf) which would allow more users to be added to channels off-chain, and liquidity to be rebalanced in these hierarchical channels off-chain. Essentially this proposal is to Lightning what Lightning is to Bitcoin, allowing more things to be done off-chain.

Second concern is channel jamming, and as far as I know there is no "good" solution to this yet. When you make a LN payment, liquidity is locked up along the route until the payment is settled. But researchers have shown channel jamming attacks to be possible, where a malicious user can refuse to settle the payment, causing liquidity to be locked up for potentially hours/days, and even trigger force-closes of channels.

As far as I know there is no good solution to the second problem yet. There are experiments of adding a "reputation score" to lightning nodes, and thus lightning nodes could limit the amount of liquidity they lock up for untrusted nodes, but this would lead to several issues. First this would lead to a somewhat worse experience for new users, because they might not be able to make moderately-sized payments until their score goes up, pushing them to centralized alternatives. Second, it's not clear if this would even work, a malicious attacker could slowly accumulate reputation and then execute the attack.

It’s actually fees that’s a concern. We saw this happen on layer 2 like polygon Chain when DeFi and NTF power users really started to run, especially at the peek.

Lightning hasn’t been tested yet, but there are concerns when things like ordinals inevitably make their way over it can get clogged - slow(er), more expensive and expensive.