Most explanations of compressed NFTs (cNFTs) lead with cost savings. That's the wrong starting point. The core design decision is moving asset state off-chain and replacing it with a single on-chain hash root — a concurrent Merkle tree root stored in an account. Everything downstream — the 1000x cost reduction, the indexer dependency, the composability friction — follows from that one architectural choice.
The cost savings are real and dramatic, but they come with a trust model shift that most documentation glosses over: your cNFT exists on-chain only as a leaf in a Merkle tree. To prove you own it, transfer it, or burn it, you need a valid proof — and that proof depends on off-chain indexers maintaining the full tree state. If every indexer goes down, the on-chain root is still there, but nobody can construct the proof needed to do anything with the asset. This is not a theoretical concern; it's a live operational dependency.