Verifying a blockchain normally means downloading and checking every transaction ever made. Light clients achieve meaningful verification while storing a tiny fraction of that data.
Full verification is expensive
A full node downloads every block, checks every signature and rebuilds the current state independently. It trusts nothing and no one.
The cost is storage measured in hundreds of gigabytes, bandwidth that runs continuously and an initial synchronization that can take days on consumer hardware.
That is reasonable for a server and unreasonable for a phone. Without a middle option, mobile users would have to trust a remote service entirely.
Headers carry commitments to everything else
Each block has a header containing a short summary of the block's contents alongside a reference to the previous block. Headers are small.
Downloading only headers produces a verifiable chain that can be checked for correct linkage and valid proof of work or attestations, at a fraction of the data.
The header does not contain the transactions themselves. It contains a cryptographic commitment that makes it possible to prove what was inside without holding it.
Merkle proofs answer inclusion questions
Transactions in a block are hashed together in pairs repeatedly until a single value remains, and that value goes in the header. The structure is a tree.
Proving a specific transaction was included requires only the sibling hashes along the path from that transaction to the root. That path grows very slowly as blocks get larger.
A light client recomputes the root from the transaction and the supplied path. Matching the header's value proves inclusion without any other block data.
The remaining assumption is about validity
Inclusion is not the same as validity. A light client can prove a transaction sits in a block without independently verifying that the block itself followed every rule.
It relies instead on the assumption that producing an invalid block is prohibitively expensive, and that honest full nodes would reject one.
That assumption is reasonable but weaker than what a full node achieves. Light clients trade a specific piece of certainty for practicality.
Proof of stake changed the design
Networks using validator attestations give light clients a different anchor, since a committee of validators signs each header and those signatures can be checked directly.
Designs that rotate a small signing committee let a client verify chain progress by checking a handful of signatures rather than examining the whole validator set.
This is why light client work matters beyond phones. Efficient verification is what lets one blockchain confirm events on another without trusting an intermediary.