An NFT Is a Token Record, Not the Asset It Points To

Ethereum’s ERC-721 proposal in January 2018 standardized a simple idea: an NFT represents a unique token record controlled by an address, not the image, song or copyright commonly shown beside it. On Ethereum-compatible networks, that record is normally identified by a contract address and token ID. The contract defines how it can be created, transferred or destroyed; everything beyond those rules depends on metadata and outside agreements.

What does the blockchain actually prove?

It proves the state recognized by the token contract. Anyone can call its ownership function, inspect its code and trace transfer events without relying on a marketplace’s account database. That evidence can establish which address controls the token and how it moved, assuming the contract behaves as expected and the underlying chain remains canonical.

The NFT’s apparent identity usually spans four separate layers:

• Token record: the contract address, token ID and current controlling address.
• Metadata: a name, description and pointer returned by the contract.
• Media: the image, audio or other file reached through that pointer.
• Rights: any license, redemption promise or access rule supplied outside the token standard.

The chain does not independently prove that a seller created the depicted work, owns its copyright or can deliver a promised physical item. Those claims require evidence from an issuer, contract or legal system.

The pointer is not the property

Metadata may live entirely on-chain, in content-addressed storage or on an ordinary web server. Those arrangements carry different failure modes. On-chain data can be costly to store but difficult to alter. A content-addressed file exposes substitution because changing the file changes its identifier, although someone must continue hosting it. A conventional URL can disappear or serve different content while the token remains untouched.

“Immutable NFT” is therefore a claim to test, not a feature to assume. Reviewers must ask whether the metadata location can change, whether an administrator can upgrade the contract and whether the media itself is retrievable. The visible artwork is often presentation; the durable object is the contract’s entry.

Who pays and who benefits?

The buyer typically pays the sale price and a network transaction fee; a marketplace may add another charge. The seller receives the proceeds, while validators or sequencers collect the network fee. Creators receive resale royalties only when the trading path recognizes them or the contract effectively enforces them. A royalty field in marketplace metadata is not a universal obligation.

Compared with a conventional database entry, an NFT can remain transferable and independently inspectable after one storefront closes. It does not eliminate intermediaries: wallets, hosting services, marketplaces, bridges and issuers still shape what holders can see or do. That separation also matters across networks, as the Manta Bridge example illustrates: the route and custody model can matter more than the asset’s familiar label.

Scarcity is enforceable; meaning is conditional

An NFT can make a particular token ID scarce and its transfers observable. It cannot make the referenced media scarce, because files can be copied, or manufacture rights the issuer never granted. Tickets, credentials, memberships and game items may gain useful portability from the mechanism, but those outcomes depend on applications continuing to honor the record.

The defensible verdict is narrow: NFTs are effective instruments for assigning and moving control of unique on-chain entries. Treating them as self-contained deeds to artwork or property mistakes a verifiable token state for the much broader bundle of claims attached to it.