In November 2025 the ratio of decentralised exchange volume to centralised exchange volume in spot crypto hit 21.2 percent, an all-time high. Read that as one in five spot dollars now trading somewhere the operator never touches your coins. Five years ago it was a rounding error. The line has only gone one way, and the reason is structural rather than fashionable.
For years the centralised exchange won on three things: custody, liquidity and user experience. It held your assets, it had the deepest order books, and it was simply easier to use than signing transactions in a wallet. That was a real moat. It is now draining, one plank at a time.
The experience gap closed
Hyperliquid is the proof. It runs an on-chain order book that feels like a centralised venue, and traders noticed. It processed $619.5 billion in perpetual volume in the first quarter of 2026 alone, and by then decentralised perps were somewhere between 60 and 70 percent of the entire on-chain derivatives market. The perp DEX to CEX volume ratio climbed from about 3 percent in January 2025 to 13 percent by December. When the hardest product to decentralise, leverage with a live order book, moves this fast, the “DEXs are too clunky” argument is finished.
Custody stopped being an asset
FTX turned the first plank from a feature into a liability. If the exchange holding your money can lose it, is lending it out, or can be frozen by a court on the other side of the world, then custody is not a service you are paying for. It is a risk you are carrying. On a decentralised exchange the coins stay in your wallet, trades settle on-chain, and counterparty risk is structural rather than something you have to trust an operator about. That is not an ideological preference. It is a cleaner risk position, and institutions are starting to price it that way.
Liquidity follows the venue, not the brand
The last plank is deepest, and it is going too. Liquidity chases volume, and volume is arriving on-chain. Market makers who once quoted only on the big centralised books now run the same strategies against DEX liquidity because that is where the flow is. As that continues, the spread advantage that justified the centralised model narrows, and the thing that kept serious size on centralised venues loses its grip.
What still has to happen
None of this means centralised exchanges vanish. Fiat on-ramps, custody for people who genuinely want a third party to hold the keys, and regulated access for institutions that are required to use a licensed counterparty are all real, and centralised venues will keep those for a long time. The point is narrower. The high-margin core of the business, matching trades and holding balances, is the exact part a well-built decentralised protocol now does at least as well, and increasingly better.
Put the two curves next to each other. Centralised share peaking and drifting down, decentralised share compounding from a low base with the fastest-growing product category already on-chain. Extend that five years and the question is not whether DEXs take meaningful share. It is which centralised exchanges reinvent themselves as on-chain venues before the flow leaves without them.
Sources: CoinGecko, DEX-to-CEX ratio; CoinGecko, State of Crypto Perpetuals 2026; CoinDesk, DEX record market share.