in the CAP theorem, every design has a weak letter.

Bitcoin is weak with C, consistency, ie, sidechains and reorgs. after 6 blocks this becomes trivial, and this is something only one PoW chain can achieve, all small chains can't build this level of momentum. Partitioning, there is the 51% attack, but that is secondary weakness because the as the network grows the percentage each new node adds is diluted progressively, and the cost of a 51% attack accelerates as the network gets larger, and more hash power comes online.

Lightning is weak with A, availability, because it depends on all nodes in a path being both online and responsive (and honest) to execute a payment.

this weakness with availability also makes it hard for the endpoints, sending is no problem, but receiving requires coordination with the sender.

this is also why the convention has been the issuance of hint-containing, route containing invoices, rather than the use of keysends, and even still... it's so delicate because of the liveness requirement and the geometric escalation of payment failure risk with the increase in network size.

but the theoretical weakness is probably overstated, same for LN as for bitcoin, because they are the first and biggest of their kind. protecting this consistency, and availability, respectively, are to the advantage of all players so the weakness in the protocol is mitigated greatly.