The state of institutional DeFi in 2026

The institutional DeFi narrative is older than the products. In 2022 every chain pitched itself as the home of institutional flow; by 2024 that pitch had been repeated so often it became background noise. In 2026 the situation is interesting because the bluster faded and a small set of credible infrastructure components are now in production. This post takes stock — what was promised, what actually shipped, what stalled, and where the surviving infrastructure is converging.

The bias is honest. We build Moby Market, so we have a stake in the answer. We try to give the rest of the field a fair reading.

What got built

Five categories of infrastructure are credible enough in 2026 that serious desks evaluate them as part of normal operations.

RFQ and quote-streaming networks. Hashflow and a handful of EVM-native peers solved one specific problem well: streaming committed quotes from professional market makers and settling against them on-chain. RFQ is now a normal swap path for mid-sized EVM flow. The model is mature; the open question is whether RFQ alone is a product or a feature embedded in larger systems.

Solver networks and intent-based execution. CoW Protocol on Ethereum and its imitators across chains established that batch auctions and intent-based execution can beat both AMMs and naive aggregators. Solver networks went from speculative in 2022 to operational in 2026, and the technical literature is maturing.

Privacy infrastructure. Zero-knowledge cryptography matured fast enough that practical privacy pools are now a normal product category rather than an academic curiosity. Renegade’s MPC dark pool, the Aztec ecosystem’s ZK rollup, Aleo’s privacy-first execution, and Solana-native ZK projects each took a slightly different tack. The common thread is that “privacy is a layer-one concern” stopped being controversial.

Cross-chain settlement primitives. Wormhole, LayerZero, and a longer tail of bridges progressed from “necessary risk” to “operational primitive”. The trust assumptions are still meaningful, but the protocols treat cross-chain as a normal expression of intent rather than as a hand-rolled script.

Execution algorithms at the protocol layer. This is the newest category and the one with the smallest credible vendor list. TWAP and VWAP as protocol-level objects, rather than as off-chain orchestration scripts that submit a stream of swaps, only became viable once chains with the throughput to run them (notably Solana) crossed the institutional usability threshold.

What stalled

A surprising number of 2022 narratives quietly evaporated.

Permissioned institutional chains. The “private chain for institutional flow” pitch did not survive contact with the institutions themselves. The institutions that wanted on-chain execution wanted it on the chains where liquidity already lived, not on bespoke environments they would have to integrate from scratch. Most permissioned-chain pitches either pivoted to L2 or quietly shut down.

Tokenised T-Bills as a flagship use case. Tokenised treasuries shipped — and now constitute a meaningful share of stablecoin reserve assets — but the narrative that they would be the entry point for institutional DeFi turned out to be inverted. Tokenised treasuries are a useful product that lives inside existing infrastructure; they did not, on their own, pull institutional execution on-chain.

“Institutional UX” as a differentiator. Every 2022 deck claimed that “institutional UX” was the unlock. In practice institutional users are far more flexible on UX than the decks assumed; they care about depth, privacy, compliance, and settlement, in roughly that order. Slick UI is a hygiene factor, not a differentiator.

KYC’d liquidity pools. Several protocols tried to differentiate via KYC’d liquidity pools — pools where every LP had to pass identity verification. The pools struggled with depth (the universe of KYC’d LPs is small) and with the political question of who runs the KYC. Most have downgraded to opt-in modules rather than mandatory pool segregation.

Where the surviving infrastructure is converging

Three convergences look durable in 2026.

Privacy is becoming a layer-one feature. Whether via on-chain ZK proofs, MPC, or rollup-level confidentiality, the assumption that institutional flow needs configurable privacy is now common across credible infrastructure. The disagreement is on the mechanism, not on the requirement. Protocols that ship without a privacy story struggle to win serious desks.

Solver networks are becoming the default execution surface. Direct AMM interaction is increasingly something only retail does. Mid-size and institutional flow goes through solver networks of one shape or another. The protocols vary on auction design, intent typing, and MEV protection mechanism, but the high-level architecture (sign intent, solvers compete, atomic settlement) is converging.

Cross-chain is becoming an intent property, not a separate workflow. The 2022 pattern was “execute on chain A, bridge to chain B as a separate step”. The 2026 pattern is to express the cross-chain requirement inside the intent itself, with the protocol handling the bridge selection and atomic settlement. Wormhole and LayerZero are the operational substrate; the user-facing primitive is the cross-chain order.

What is still genuinely unsolved

A few problems remain hard enough that no project has a fully satisfying answer.

MEV under adversarial solvers. Solver networks help by aligning solver incentives with user outcomes, but a sufficiently sophisticated adversarial solver can still extract value from the batch boundary or the commit-reveal window. The frontier is in better cryptographic enforcement (threshold encryption for batches, more aggressive slashing on misbehaviour) and more transparent post-trade audit.

Cross-chain trust assumptions. Wormhole and LayerZero are operational, but neither offers cryptographic guarantees as strong as same-chain settlement. The trade-off is real and gets reflected in lower acceptable trade sizes for cross-chain orders. Better bridge designs (light-client-based, ZK-bridged, restaking-secured) are in development but not yet a defaults.

Compliance integration. Selective disclosure is a real cryptographic primitive but the regulatory acceptance of it lags the technology by years. Desks that need explicit regulator buy-in still rely on traditional KYC and on-chain identity hooks more than on ZK assertions, even though the assertions are technically stronger. This is a market-development problem more than a technology one.

Liquidity fragmentation. The proliferation of L2s and app-chains fragmented liquidity in ways that even good cross-chain primitives only partially solve. The 2024 dream of a single unified liquidity surface is still ahead of us; the 2026 reality is that desks operate across many venues with explicit per-venue depth modelling.

Where Moby Market sits in the landscape

We built Moby Market because the convergences above looked clear enough by 2024 to be worth committing to. The bet is that:

1. Privacy and execution-algo semantics belong in the same protocol surface, not as separate products.
2. Solana’s throughput makes protocol-level execution algos economical in a way that EVM cannot match without significant compromise.
3. Cross-chain settlement as an intent property is the right interface, with Wormhole and LayerZero as the substrate.

Those bets are not unique to us; they are the bets the credible infrastructure is making. Moby Market’s distinctive choice is to integrate the bets into a single Rust workspace with a coherent type system, rather than letting them live as separate composed protocols.

The risk of that choice is integration complexity and a smaller TAM than a focused single-feature protocol would have. The opportunity is a coherent execution surface that a single desk can adopt without negotiating between five vendors. We will know in eighteen months whether that bet was right; today the early production usage suggests it is at least defensible.

Predictions for the next eighteen months

Calibration is hard but worth attempting. Our bets:

• Solver network consolidation. The long tail of solver protocols thins to roughly three credible networks per chain ecosystem. Solver economics favour scale and good auction design has converged enough that further differentiation is marginal.
• Privacy becomes table stakes. By 2027, infrastructure protocols without an on-chain privacy story will be hard to evaluate for serious desks. The expectation will be “of course there is a privacy layer” the same way TLS is now expected on any public web service.
• Cross-chain trust models stratify. Cheap bridges will continue for retail flow; institutional flow will move to stronger trust models (light-client, ZK-bridged, restaking-secured) once those mature. Wormhole and LayerZero will probably remain operational substrate but with stronger trust-extension layers on top.
• Tokenised assets pull liquidity, not narrative. The next eighteen months will see meaningful institutional flow tied to tokenised real-world assets, but the framing will be “asset migration to better infrastructure” rather than “institutional DeFi”. The narrative will recede; the volume will rise.
• Compliance lags by another two years. Selective disclosure and other ZK-based compliance primitives will not gain mainstream regulatory acceptance inside the prediction window. Desks will continue to use traditional KYC alongside the cryptographic primitives.

The takeaway

Institutional DeFi in 2026 is unglamorous in a healthy way. The hype receded; the surviving infrastructure works; the design choices that matter are increasingly clear. The next eighteen months will reward protocols that compose the established primitives well and punish those that try to relitigate questions the field has already answered.

We have a stake in that outcome. So do most of the protocols we compare ourselves to. The right read on the field is not zero-sum competition — it is shared infrastructure for a market that is finally large enough to support more than one credible builder.