CEXs, exchange-backed chains and white-label markets
Centralized exchanges did not centralize because they failed to understand blockchains. A shared account system, custody arrangement and matching engine can remove a great deal of coordination from the user's day. Speed, customer support, fiat access and concentrated liquidity are real benefits. So is account recovery when it works.
The useful comparison is about authority: who holds assets, chooses what users see, processes orders, and can change the rules? A wallet-connect button does not answer those questions.
Efficiency involves delegation
On a typical custodial CEX, a trade first changes the operator's internal ledger. It need not wait for a separate public-chain transaction. Netting and common margin arrangements can reduce friction. Actual costs still depend on depth, fees and operating quality; CEX is not a synonym for cheap execution.
The same integration creates exposure to the operator's finances, custody, availability and decisions. A displayed balance is not independent evidence of complete liabilities or withdrawal capacity. Proof of reserves also needs careful interpretation: the SEC's investor bulletin explains why such reports are not equivalent to financial-statement audits. Investor.gov
A noncustodial DEX can replace some institutional promises with inspectable execution rules. It introduces other burdens: contract risk, gas, slippage, approvals, chain congestion, MEV and key management. For some users, a capable regulated service remains a reasonable choice.
| Question | Typical custodial CEX | Noncustodial contract DEX |
|---|---|---|
| Where does a trade settle? | Internal ledger; withdrawal is separate | Protocol changes on-chain state |
| Who can move assets? | Custody arrangement controlled by operator | User authorization and contract rules |
| What supports execution? | Concentrated orders, inventory and services | Pools, signed quotes and composable protocols |
| What can recover a mistake? | Support and operator intervention may help | Often limited; some protocols retain admin powers |
| What must be trusted? | Business, finances, custody and controls | Contracts, chain, approvals and access infrastructure |
Hybrid designs need a more detailed analysis than this table.
An exchange-backed chain is not an internal ledger
Base, incubated by Coinbase, is an Ethereum L2. Its Stage 1 announcement describes permissionless fault proofs and broader upgrade approval. Those are substantive changes, regardless of corporate origin. But permission to deploy an application, verify state, order transactions and authorize upgrades are separate powers. A single milestone cannot certify every layer. Coinbase's neutrality principles separately address sequencing and nonpublic transaction data. Base announcement, principles
BNB Smart Chain uses PoSA and a staking-based active validator selection process, with an epoch block-producing subset. Its relevant questions include validator access, stake concentration, governance and incident coordination. Transactions on that chain are not simply entries in Binance's exchange database; an application deployed there still needs its own contract and authority analysis. Validator documentation
The original Binance DEX is a useful historical counterexample to brand-based classification. Its 2019 launch described noncustodial trading on the then-native chain. That does not establish today's service status, and the original Binance Chain should not be conflated with later BSC architecture. 2019 launch
A CEX app can also expose a third-party DEX. The user may retain keys while depending on that app's rankings and routing. Follow the particular transaction: account or wallet, whose signature, whose route, which settlement rules?
White-label is a delivery model
White-label software lets an operator launch a branded service with supplied engines, accounts, administration and integrations. AlphaPoint's brokerage offering, for example, describes institutional control over accounts, pricing, execution and exposure. That reduces the cost of operating a business; it does not give each customer a market node. Product description
White-label software could also use noncustodial settlement. The label alone says nothing conclusive about custody. Ask whether orders depend on one backend, whether another interface can complete them, whether identities and data can move, and what happens if the vendor disappears.
A hundred logos sharing an irreplaceable backend do not establish a hundred independent markets. Letting anyone open an exchange expands the supply of operators. Letting individuals own their state, terminal and agent changes the capabilities of participants. That second possibility is the one this series investigates.