People use "coin" and "token" as if they mean the same thing. Most of the time, the confusion doesn't cause real problems. But there's a genuine architectural difference between the two — and that difference has consequences for custody, risk, and regulation that are worth understanding clearly.
The short version: a coin is the native asset of a blockchain. A token is an asset that lives on top of one.
That sounds almost too simple, so let me show you why it matters.
A coin is issued and maintained at the protocol layer. It exists because the blockchain itself was designed around it. The network uses it to pay for computation, reward validators or miners, and in many cases, to secure the chain economically.
Bitcoin (BTC) is the clearest example. The Bitcoin network was built specifically to produce and transfer BTC. There is no contract governing BTC — it's built into the Bitcoin protocol itself. Miners receive BTC as a block reward; users pay BTC fees to get transactions included. The coin and the chain are inseparable.
Ethereum (ETH) works the same way. ETH is used to pay gas fees, and since The Merge, it's also staked by validators who need it at risk to participate in consensus. There's no ERC-20 contract for ETH — it's the chain's native asset, not an application running on top of it.
Other examples: SOL on Solana, ADA on Cardano, AVAX on Avalanche, BNB on BNB Chain. Each of these is the native currency of its respective network.
A token is a different thing architecturally. It's an asset defined and governed by a smart contract deployed on an existing blockchain. The contract specifies how many tokens exist, who can create or destroy them, and what the rules are for transfer.
USDC is a token. Circle wrote and deployed an ERC-20 contract on Ethereum that defines USDC's supply and transfer logic. The actual settlement happens on Ethereum — ETH pays for the gas — but USDC itself is just state tracked inside Circle's contract. When you send USDC, you're calling a function in that contract that decrements one balance and increments another.
The same logic applies to DAI (controlled by MakerDAO's smart contracts), UNI (Uniswap's governance token), WBTC (wrapped Bitcoin, an ERC-20 backed 1:1 by BTC held in custody), and most of what people call "DeFi tokens." These are all contracts running on Ethereum's infrastructure. They use ETH to pay for their own execution.
Tokens exist on most major chains, not just Ethereum. Solana has SPL tokens (governed by the SPL Token Program). BNB Chain has BEP-20s. Avalanche has its own token standard. The underlying principle is the same across all of them.
The distinction isn't aesthetic. It has three concrete implications.
Issuer risk. Coins don't have issuers in the traditional sense — the protocol mints them according to fixed rules. Tokens do have issuers. Circle controls the USDC contract; it can blacklist addresses, freeze funds, and upgrade the contract. If the issuer is compromised, or the contract contains a bug, the token can behave in ways the holder didn't expect. The chain itself might be fine while the token is broken.
Smart contract risk. A token's behavior is entirely determined by its contract code. If that code has a vulnerability, it can be exploited — and token exploits are distinct from chain exploits. The Ethereum blockchain didn't fail in the DAO hack; a specific contract did. The coin (ETH) was unaffected at the protocol level, even as the token (DAO governance tokens) collapsed.
Upgrade authority. Many tokens are governed by upgradeable contracts with admin keys. This is often disclosed but frequently overlooked. An upgradeable token contract means whoever holds the admin key can change the rules. Some tokens have time-locked governance processes around upgrades; others don't. This varies enormously and isn't visible from the token's surface appearance.
For coins, the constraints are at the protocol level — changes require network consensus, hard forks, and significant coordination. Bitcoin's 21 million supply cap isn't a number in a contract someone could update; it's wired into the consensus rules every full node enforces. That's a hard constraint.
For tokens, constraints are softer. They live in contract code that may or may not be immutable. The Tether (USDT) contract has a function to freeze addresses — that's a soft constraint imposed by the issuer, not the chain. MakerDAO's DAI is governed by on-chain voting, which is decentralized but still a process that can change parameters like stability fees or collateral types.
The practical takeaway: coins are constrained by protocol design. Tokens are constrained by contract code and whoever controls it.
A few structural shifts are worth tracking.
The regulatory distinction between coins and tokens is increasingly active. The SEC's position — contested and still evolving as of early 2026 — is that many tokens meet the criteria of the Howey Test (investment of money in a common enterprise with expectation of profit from others' efforts), which would classify them as securities. Coins, particularly Bitcoin, have generally been treated as commodities. Ethereum's classification has been debated. This isn't resolved, but enforcement actions continue to define the boundary by exclusion rather than bright-line rule.
On the technical side, the ERC-4626 standard (vault tokens with standardized yield-bearing logic) is creating a new sub-category of tokens with built-in yield mechanics. This has meaningful DeFi composability implications but doesn't change the underlying coin/token architecture.
BNB is worth noting as a historical case of the boundary moving: BNB started as an ERC-20 token on Ethereum in 2017, then migrated to become the native coin of BNB Chain (launched 2020). That transition changed it from a token to a coin — a migration that's possible but rare, and that required Binance to build an entirely new L1 chain to accomplish it.
The core mechanism remains stable. The distinction between protocol-native assets and application-layer assets isn't going away.
For the regulatory dimension: a clear legislative or judicial ruling that definitively categorizes specific tokens as securities (or explicitly exempts them), stable enforcement patterns across multiple jurisdictions, and institutional custody providers standardizing their classification processes accordingly.
For the technical dimension: if a major token migration (token → native coin) happened on a significant chain, that would be worth noting. If account abstraction (ERC-4337) evolved to blur the distinction between coins and tokens meaningfully at the user layer, that would warrant updating the analysis.
The architectural distinction is unlikely to break, but it could become less practically relevant if dominant chains adopted account abstraction so thoroughly that the distinction between gas-coin and application-token became invisible to users — or if regulatory treatment converged to treat coins and tokens identically, removing the practical risk differential.
Neither of these is imminent.
Now: The regulatory line between coins and tokens is live and affecting institutional decisions — exchanges, custodians, and funds are making real choices based on how assets are classified. Understanding the architectural difference is relevant for anyone making custody or risk decisions today.
Next: ERC-4626 and related token standards are actively evolving. Worth tracking if you're interacting with DeFi protocols in a serious way.
Later: Full account abstraction at scale could eventually reshape how users experience the coin/token distinction — but that's still years from mainstream.
This post explains the architectural distinction between coins and tokens. It doesn't cover every token standard or chain-specific implementation in full detail, and it doesn't address tax treatment, which varies by jurisdiction and classification.
Nothing here constitutes investment advice or a recommendation to hold any particular asset.
The distinction is structural. What you do with it is a separate question.




