Ethereum's Aggregator Split Is a Warning for Passive LPs
The most important liquidity story this week is not another pool launch or another fee switch vote. It is a quieter routing change on Ethereum.
On April 15, 2026, The Block reported that Ethereum's DEX aggregator market has become much less concentrated: Kyber leads with about 31% direct aggregator share, CoW Swap follows around 22%, and 1inch has fallen from roughly 30% to 15% over the same period (The Block).
For LPs, this is a warning that "where the volume goes" is no longer something you can infer from pool depth, protocol brand, or historical dominance. Routing is becoming its own competitive layer, and that layer can redirect order flow faster than most passive LPs can react.
Aggregator Share Is Not Just UI Share
DEX aggregators are not just interfaces. They are the attention layer, routing layer, gas-cost layer, MEV-protection layer, and sometimes the solver layer sitting above AMMs. When their share changes, it changes which pools get touched, which fee tiers matter, which quotes are visible, and which LPs actually receive flow.
The Block's framing is useful because it notes two things at the same time: the Ethereum aggregator market is more competitive than it has been in years, and the direct-to-aggregator data does not capture every backend routing relationship. If a trade starts at one aggregator and routes through another system underneath, attribution can miss part of the real influence (The Block).
That caveat is the point. The visible winner may not be the only venue shaping execution.
Dune's own DEX data docs show why this category matters: aggregator trades are tracked separately from raw DEX trades, alongside MEV and sandwich-attack datasets, because routing paths are now part of the execution market rather than a minor detail (Dune docs).
The Old LP Shortcut Is Breaking
The lazy LP model used to be simple:
- find the pool with the deepest liquidity,
- assume routers will prefer it,
- pick a fee tier,
- and wait for volume.
That still works sometimes. It is also becoming a weaker assumption.
If aggregators are competing on routing quality, gas, solver performance, MEV protection, private order flow, or rebates, the deepest pool is no longer automatically the best destination for every trade. A route can split across multiple pools. A solver can internalize flow. A front end can prefer an execution path that looks worse in raw pool depth but better after gas, slippage, failed-transaction risk, or MEV protection.
This is why I do not view 1inch losing share as mainly a 1inch problem. It is a signal that LPs should stop treating historical aggregator dominance as a stable source of demand. If incentives no longer explain all the flow, what does? Better fills? Better UX? CoW-style batch auctions? Kyber's routing surface? Better wallet integrations? Market makers preferring one settlement style over another?
Each answer implies a different LP risk.
CoW Flow Is Different Flow
CoW Swap is not just another pathfinder pointing at AMM pools. Its model is built around intents, batch auctions, and solver competition. That means some trades can be matched against other trades before touching AMM liquidity at all.
If two users can be crossed directly, the AMM pool does not earn the fee. If a solver finds a better path through a mix of inventory, private liquidity, and on-chain venues, the pool becomes one execution option rather than the default counterparty. If more users choose intent-based execution because they care about MEV protection and price certainty, then LPs are competing not only with other LPs, but with non-AMM liquidity.
This does not make AMMs obsolete. It makes them less guaranteed.
The best pools will still win flow when they offer the best net execution. But "best" now includes more variables than price impact at the pool level: gas, latency, MEV leakage, solver inventory, RFQ-style liquidity, and whether a route can avoid public-mempool toxicity.
Kyber's Lead Says Routing Still Has Alpha
Kyber leading direct Ethereum aggregator share is the other side of the story.
If the market were already fully commoditized, it would be hard for a routing venue to pull ahead this visibly. The fact that Kyber can sit around the top of direct aggregator share suggests routing quality still has meaningful variance. For traders, that is good. Competition between aggregators should reduce complacency and force better execution.
For LPs, it is more complicated.
Better routing is not the same thing as better LP returns. A stronger router can discover cheaper liquidity, split order flow more efficiently, and route around pools whose fee tier is too expensive. That may improve trader outcomes while compressing the easy fee capture that passive LPs expected.
This is the hidden cost shift: the more competent the routing layer becomes, the less LPs can rely on being conveniently nearby. Your pool has to be useful on the margin. Otherwise, it is just displayed liquidity that gets skipped when the route is optimized.
The Pool Is No Longer the Whole Venue
This is the part DeFi still underestimates.
A DEX pool is not a venue by itself anymore. The venue is the stack:
- wallet or front end,
- aggregator,
- solver or pathfinder,
- MEV protection layer,
- AMM pool,
- fallback route,
- and settlement chain.
LPs own only one piece of that stack. They provide inventory at a price curve. They do not control who sees it, who routes to it, how toxic the order flow is, or whether an intent system bypasses it.
That does not mean LPing is a bad business. It means LPing is now closer to market making under fragmented broker routing than passive yield farming. If your position stops earning, the cause may not be the pool itself. It may be that an aggregator changed behavior, a solver found better inventory elsewhere, a wallet integration shifted flow, or a competing route made your fee tier less attractive.
What LPs Should Watch Now
The practical takeaway is not "use Kyber" or "avoid 1inch" or "CoW will eat AMMs." The better takeaway is that LPs need to watch routing share as a first-order variable.
For an active LP, I would track:
- whether volume is coming from aggregators or direct pool users,
- whether the pool is selected during volatile periods or only quiet periods,
- whether fee income improves when routing competition increases,
- whether MEV-protected or intent-based flow is bypassing the pool,
- and whether a pool's TVL is deep but unused.
The last one is especially important. Deep unused liquidity is not a moat. It is idle inventory.
In a more fragmented aggregator market, LPs should care less about being in "the biggest pool" and more about being in the pool that routers keep choosing when execution quality actually matters.
My Take
The Ethereum aggregator split is healthy for traders and uncomfortable for passive LPs. It means the routing layer is alive, execution quality is still contested, and a single historical winner does not own the order-flow map forever.
But it also means LPs are underwriting a moving distribution system. You can be in a technically sound pool, with reasonable depth and a sensible fee tier, and still see fees disappoint because the route selection layer moved around you.
That is the real market-structure lesson from Kyber and CoW gaining share while 1inch loses ground:
liquidity does not get paid for existing. It gets paid when the execution stack decides it is useful.
For LPs, that is the metric to watch now. Not just TVL. Not just APR snapshots. Not just protocol brand.
Watch who controls the route.