Canton doesn't need another AMM. It needs a dealer market.
The short version
On a network designed for institutional assets and participants, Meriq’s view is that liquidity can be quoted competitively by independent market makers and settled atomically, without requiring all capital to sit behind one shared pricing curve.
Ask most people why onchain trading uses automated market makers and you'll get some version of "that's just how DeFi works." It's treated as a law of physics rather than a design choice made under specific constraints, at a specific moment, for a specific reason.
Those constraints were real. Early DeFi had to build markets between strangers, with no credit relationships, no compliance layer, and no dealer willing to extend balance sheet to an anonymous wallet. A fully onchain order book was too expensive to run. So the industry did something clever: it put inventory in a smart contract and let a formula set the price. No dealer, no counterparty risk, no trust required. For the environment DeFi built for itself, the AMM was close to the right answer.
Canton is not that environment.
It's built for institutional assets, known counterparties, privacy, and atomic settlement. Most of the firms expected to bring liquidity to it are already market makers somewhere else, on Nasdaq, in FX, in the OTC crypto desks that never touched an AMM in their life. They already know how to manage inventory and price risk. The question isn't how to wrap an AMM around every Canton asset. It's what liquidity looks like when you let those firms do what they already do, just onchain.
Often, that looks like a dealer market.
Two different jobs
An AMM commits capital before anyone shows up. Deposit into the pool, let the curve decide the price. The liquidity exists whether or not a trade ever arrives.
A dealer waits for the trade, then decides. Does it want the risk? At what price? That price depends on more than the asset; it depends on the dealer's whole book that day.
Say four firms see the same order. One is overweight the asset and wants out. One wants to build a position. One has cheap financing and can hedge for nearly nothing. One is just tapped out on risk that week and quotes wide, or doesn't quote at all. Four different prices, for rational reasons. A pool can't do that. A pool has one curve and one answer, forever, regardless of who's asking or why.
That's not a bug in dealer markets. It's the whole point. Price isn't just a function of the asset; it's a function of who's willing to hold it right now. This is particularly true for something like a tokenized Treasury, and the dealer is pricing maturity, funding cost, redemption terms, and the cash market alongside the token itself. That's ordinary sell-side work. Nothing about putting the trade onchain makes that work obsolete.
What Canton actually changes
None of this is new: dealers, competing quotes, inventory-based pricing. What's new is doing it onchain without blowing up the privacy institutional players require.
A market maker on Canton doesn't have to publish its book to the whole network to compete for flow. A trader doesn't have to broadcast an order to a public mempool before it executes. And once two sides agree on a price, the trade still settles atomically: no partial fills, no counterparty risk, no waiting for a rollup to finalize.
Early DeFi collapsed pricing and settlement into a single object: the contract held the inventory, set the terms, and settled the trade, all at once. Given the constraints at the time, that was the right call; it replaced the dealer with code because there was no dealer to work with. Canton removes that constraint. It lets you split the job back into its two natural parts: a market maker decides what it's willing to risk, and the network makes the resulting agreement final. Each side does what it's actually good at.
This isn't hypothetical. It's already happening
0x's RFQ system and UniswapX both let market makers compete to fill an order instead of forcing every trade through a pool. Settlement still happens onchain. Pricing comes from firms running their own models and holding their own inventory.
AMMs didn't fail. They're still the right tool for a lot of markets, and will be for a while. But the fact that RFQ systems exist at all inside AMM-native ecosystems tells you something: even there, the line between "how a trade settles" and "who provides the price" is getting less rigid, not more. On Canton, that line was never rigid to begin with. Market makers aren't a bolt-on feature request. They're the participants the network was designed around.
TVL was always the wrong number
Crypto latched onto TVL because it's visible: a dashboard number, easy to screenshot, easy to compare. But no trader actually cares how much capital is sitting in a contract. They care what price they get, at the size they need, right now.
A dealer market answers that question directly: multiple firms look at the same order and compete on it, live, based on their own risk and inventory. A dealer might comfortably fill a large order it never "committed" any capital to, simply because the trade happens to net down its existing book. That liquidity was real, it was just invisible five minutes earlier, sitting on someone's balance sheet instead of locked in a pool.
For a network that's going to host dozens of tokenized asset types, pre-funding deep pools for every pair is a bad use of capital. It's often smarter to route the trade to a balance sheet that can price it the moment it shows up, rather than a pool that was funded, or underfunded, in advance.
Where Meriq fits
That's the model behind Meriq. Submit a swap, get an executable price. Independent market makers compete for the flow, each keeping full control of its own inventory and risk. Meriq routes to whoever's pricing best, and the trade settles atomically on Canton.
No shared pool. No common curve. Every market maker on the other side of a trade made an independent decision to take that specific risk, at that specific price, which is exactly why the price is good.
AMMs aren't going away on Canton, and some assets will suit pooled liquidity fine. But onchain liquidity and pooled liquidity have quietly become synonyms in most people's heads, and they shouldn't be, especially on a network built for institutional assets and institutional participants, not anonymous liquidity providers. The AMM solved DeFi's original problem. Canton has a different one. It's worth building the market structure that actually fits it.