The problem that needs to be solved is, basically, that the fundamental architecture of Bitcoin establishes an upper limit on transactions-per-second (TPS) which is too low.

Lightning "solves" this problem by moving transactions off-chain, bundling them into batches, and then emitting batches of transactions to Bitcoin mainnet. That means a transaction against Lightning is only valid once the bundle in which the transaction is placed is committed on mainnet. A confirmation from Lightning isn't good enough; users have to wait for a confirmation from mainnet.

Batching is a common optimization! But it's not a solution to performance problems. It weakens the guarantees afforded by transaction confirmation. Those guarantees are important. They define the consistency model provided by the system. The model of Bitcoin which includes Lightning is fundamentally different, and weaker, than the core model of Bitcoin.

And so Lightning isn't a solution to the problem. It weakens the transactional model of Bitcoin, and so provides a totally separate system to Bitcoin proper.