Reusing capital instead of duplicating it
In a standard Proof of Stake system, validators lock up capital to secure one chain. That stake backs one set of rules, and if a validator misbehaves on that chain, the stake is slashed. Restaking starts from a simple observation: that same locked capital represents real economic value at risk, and there's no fundamental reason it can only ever back one thing at a time. If a validator is already trusted enough, and has enough capital on the line, to secure a base chain, that same capital could in principle also stand behind other services, as long as there's a mechanism to hold it accountable for misbehavior in those services too.
Restaking is the protocol layer that makes this reuse possible. Rather than a new network having to convince a fresh set of participants to acquire and lock up entirely new capital, it can instead ask validators who already have capital staked on an established base chain to opt in, extending their existing stake to also secure the new service. The capital doesn't move or get duplicated, it gets additional obligations layered onto it. This is the core shift: security stops being something each network builds independently from zero, and becomes something that can be borrowed, in a cryptoeconomically enforced way, from capital that's already doing productive security work elsewhere.