The Cosmos Chains That Matter: Osmosis, Neutron, and Stride

Three bets, one ecosystem

Part 1 of this series argued that a sovereign appchain beats being a tenant on someone else's chain: no gas auctions, no competing for block space with a thousand unrelated applications, full control over your own execution environment. Part 3 covered how the Cosmos Hub tried to turn that sovereignty into a subscription business through Interchain Security (ICS) — renting out ATOM's validator set to other chains for a fee.

This part is the practical tour. Four chains, four different answers to the same question: how much of your own security do you want to own, and what do you actually build once you own your chain?

Say the quiet part up front: none of this is big. Not by Ethereum's standards, not by Solana's. The numbers in this article stay small all the way through, and that's not the interesting part. The interesting part is that these appchains, built on variations of the same Cosmos SDK toolkit, made genuinely different strategic choices — in public, with consequences — and those choices produced real product differences that don't exist when you're a contract deployed on somebody else's shared chain. Scale and whether the model works are two separate questions. This article treats them separately.

Osmosis: the DEX that owns its chain

Osmosis is a Cosmos SDK chain built for one purpose: being an automated market maker (AMM) — a decentralized exchange that prices trades algorithmically against pooled liquidity instead of matching individual buy and sell orders like a traditional order book (see how an AMM actually prices a trade). Everything else Osmosis does is downstream of that one job.

What makes Osmosis worth studying isn't that it's a DEX — Cosmos has other DEXes. It's that Osmosis controls the entire chain the DEX runs on, not just a contract deployed on somebody else's. That distinction produces features that can't exist as a mere contract:

• ProtoRev. Built with Skip, this is a protocol module — code that runs as part of block production itself, not a deployed contract — that detects cyclic arbitrage opportunities created by ordinary user swaps at the end of every block, flash-mints OSMO to capture that arbitrage atomically, and burns the profit or routes it to destinations set by governance. In practice, MEV (maximal extractable value — profit that would otherwise leak to outside bots and searchers) gets captured by the protocol instead of by external arbitrageurs. By the first half of 2025, ProtoRev had burned more than 3.2 million OSMO this way.
• Superfluid staking. A liquidity provider can simultaneously stake the OSMO half of their LP position to a validator — earning swap fees and incentives from the pool and staking rewards from the same locked capital.
• Concentrated liquidity, added in 2023, lets LPs deploy capital into a chosen price range instead of spreading it across the entire 0-to-infinity price curve. Osmosis documents this as delivering roughly 200–300x higher capital efficiency than full-range provision — though concentrating capital into a narrower band also concentrates the risk of impermanent loss when price moves outside your chosen range.
• More recently: "alloyed assets," unified representations of the same underlying BTC or stablecoin that arrived over different bridges, and the Polaris Token Portal, a cross-chain trading interface launched in September 2024.

None of that requires Osmosis to be big to be real. Consensus-level MEV redistribution and native staking-LP composability are structural features of owning the chain. A contract on someone else's chain doesn't get to touch block production.

Osmosis tokenomics: from 73% APY to near-zero

OSMO's original emission schedule was aggressive by design: year-one issuance implied staking returns above 70% APY, stepping down on a "thirdening" schedule, where each period's emissions run roughly a third lower than the last. That was a bootstrap mechanism, not a sustainable yield.

"OSMO 2.0," reported in June 2023, cut inflation by roughly half, targeting around 11%, and added two things: fee-sharing that routes protocol revenue to stakers, and a burn mechanism aimed at eventually pushing OSMO toward net deflation. The next scheduled Thirdening took effect in June 2025, cutting annual emissions from 9% to 6%.

The result: OSMO staking yield today sits around 2% APY. That's a genuine climbdown from the launch design, and an honest one — Osmosis chose to trade a hyperinflationary bootstrap for a sustainable long-term supply curve, years ahead of when most token launches make that adjustment.

Osmosis's honest scale check

Here's where the modesty of Cosmos DeFi becomes unavoidable. Osmosis TVL (total value locked — the dollar value of assets deposited in a protocol's pools and contracts) peaked near $1.83 billion in March 2022, crashed to roughly $316 million by that May, partially recovered to somewhere around $150–180 million by late 2024, and as of 12 July 2026 sits at roughly $13.75 million. Daily DEX swap volume runs in the low single-digit millions, around $1.2–1.7 million a day.

Osmosis deserves credit where it's due: it processed $2.75 billion in trading volume in the first half of 2025 alone, more than $41 billion cumulatively since launch. Volume and TVL tell different stories — a lot of capital has passed through Osmosis even as the amount sitting in its pools at any given moment has shrunk.

Context matters here too. The total crypto market cap fell from a $4.2 trillion all-time high in October 2025 to roughly $2.2 trillion by mid-2026, a decline of about 48%. But Cosmos DeFi's decline has clearly outpaced that broader market drop. Osmosis is the biggest fish in a small pond: dominant within Cosmos DeFi, not a competitor to EVM (Ethereum Virtual Machine) DeFi at large.

Neutron: CosmWasm's bet on full sovereignty

Where Osmosis is a single-purpose AMM chain, Neutron is general-purpose: a CosmWasm smart-contract platform, meaning developers deploy contracts onto it much like they would onto Ethereum, except the chain itself is Cosmos SDK-based and settles over IBC (Inter-Blockchain Communication) with the rest of Cosmos.

Neutron's origin story matters more than most. It launched in 2023 as the Cosmos Hub's first Interchain Security consumer chain — the literal flagship proof-of-concept for the ATOM monetization model covered in Part 3. Neutron didn't build its own validator set. It borrowed the Hub's.

Then it left. The "Mercury Upgrade" went live on 9 April 2025, migrating Neutron off Hub-secured replicated security and onto a fully sovereign proof-of-stake network with its own validator set. The stated reasons: the ICS arrangement wasn't creating meaningful value accrual for ATOM, it placed a heavy operational burden on the Hub's validators, and — the part that stung most for a fast-moving DeFi chain — it slowed Neutron's own upgrade cadence. Being secured by someone else's validators means moving at someone else's pace.

Neutron's replacement is a DAO-curated validator set of roughly 20 validators, with NTRN token supply earmarked to fund that set's operation over the next couple of years. Worth noting alongside the security exit: in December 2023, Neutron took a 25% ownership stake in Confio, the company behind the CosmWasm framework itself — a vertical-integration move to keep CosmWasm's own roadmap aligned with Neutron's needs.

Neutron's apps and honest TVL

Neutron's flagship applications show what a general-purpose sovereign chain can do that a single-purpose one can't. Astroport is a Curve-style AMM, optimized for low-slippage swaps between similarly priced assets like stablecoin pairs. Duality is Neutron's own native, in-protocol order book module — built into the chain itself rather than deployed as a contract, the same "own the base layer" move Osmosis makes with ProtoRev. Drop Protocol handles liquid staking, issuing dATOM. On top of these sit vault and market-making strategy builders that compose across all of them.

The concrete composability example: Astroport integrated Passive Concentrated Liquidity pools with Duality's native order book, so a single trade can route across both a deployed AMM contract and a native chain module in one atomic transaction. That's the appchain thesis working as advertised — a level of interoperation between "the chain itself" and "a contract on the chain" that doesn't exist when you're a tenant.

Scale, honestly: Neutron's chain TVL sits at roughly $5.57 million as of 12 July 2026.

Stride: liquid staking, and the opposite ICS story

Stride is an appchain purpose-built for one job: liquid staking. Deposit ATOM, TIA, OSMO, or DYDX, and Stride stakes it with validators on your behalf while minting a liquid staking token (LST) — stATOM, stTIA, stOSMO, stDYDX — that represents your staked position and keeps earning staking rewards, while remaining usable elsewhere in DeFi as collateral or a trading asset. The problem an LST solves: normally staked tokens are locked and illiquid; an LST gives you both the yield and the liquidity.

Here's the twist this article is built around. Stride went the opposite direction from Neutron. On 19 July 2023, Stride formally adopted Interchain Security, handing block production over to the Cosmos Hub's validator set specifically to inherit the Hub's economic security. That move took Stride's economic security from roughly $25 million, when secured by its own STRD-staked validator set, to about $2.3 billion, secured by ATOM. Overnight.

That is a genuinely good reason to rent security. A liquid staking protocol's entire business model depends on being trusted with other people's staked capital — a validator set compromise or chain halt is existential for it in a way it isn't for, say, a DEX that can recover and keep trading the next day. Borrowing $2.3 billion of security was worth far more to Stride than it turned out to be worth to Neutron.

As of the most recent confirmable evidence — a September 2024 Cosmos Hub governance forum thread renegotiating validator compensation for consumer chains — Stride remained an active ICS consumer chain. No exit has been documented, in clear contrast to Neutron's well-publicized departure seven months later.

Scale: Stride's protocol TVL is roughly $5.58 million as of mid-2026. Liquid staking did find real traction within Cosmos — Stride holds a dominant share of Cosmos LSTs — but "real traction" here still tops out at single-digit millions of dollars.

Put the two chains side by side and the lesson is clean: ICS didn't simply succeed or fail. It fit one chain's profile and not the other's.

Neutron was a fast-iterating DeFi hub that found ICS's upgrade drag intolerable. Stride is a staking-derivative issuer whose whole product is a trust relationship with depositors' capital, so borrowing enormous economic security was worth the tradeoff. Same starting model, opposite conclusions, both defensible.

dYdX: the chain that actually left Ethereum

If you want the single strongest piece of evidence for the appchain thesis, it isn't any Cosmos-native chain — it's the one that came from outside and left Ethereum's orbit entirely.

dYdX v3 ran as a perpetuals exchange — a derivatives product letting traders take leveraged long or short positions with no expiry date — with an off-chain order book matched against on-chain settlement, built on StarkEx, an Ethereum layer 2. That's a completely different lineage from Osmosis, Neutron, and Stride: dYdX had no Cosmos history and no reason to move except that the appchain model offered something Ethereum's L2 stack couldn't.

dYdX Chain — v4 — launched its genesis mainnet on 26 October 2023, rebuilding the exchange from scratch as a sovereign Cosmos SDK appchain with its own validator set. This is the strongest appchain proof point precisely because it was a deliberate exit by a team with zero prior Cosmos lock-in, choosing sovereignty over staying inside Ethereum's ecosystem.

Where does that leave dYdX in 2026? It's the second-largest decentralized perpetuals exchange by trading volume — but Hyperliquid, a purpose-built, non-Cosmos layer 1 designed around the same problem, now captures more than 70% of decentralized perp volume. dYdX's chain TVL sits at roughly $89.8 million as of 12 July 2026, down from about $221.7 million a year earlier — a steep decline. But that number is still larger than Osmosis, Neutron, and Stride combined, making dYdX the largest single appchain economy in this article by a wide margin.

The honest verdict: dYdX succeeded at exactly what the appchain thesis promised — sovereign infrastructure, no gas-fee competition with unrelated applications, real product-market fit at genuinely material scale. It's just no longer winning the category it helped pioneer, as the broader decentralized perps market has moved toward Hyperliquid instead.

The scale check

Numbers, plainly, mid-2026 chain or protocol TVL:

Add up the named Cosmos appchains here and total DeFi TVL across them is roughly $120–150 million — under 1% of Ethereum's, under 3% of Solana's. Ethereum's own dominance share of DeFi TVL has been sliding too, from about 63.5% at the start of 2025 to around 54% by May 2026, but that's a story about Solana and other large chains taking share, not about Cosmos.

One line in that table deserves to land quietly and hard: the Cosmos Hub itself — the chain that Interchain Security was designed to enrich by renting out its validator set — sits at roughly $126,000 in TVL. Essentially nothing. All the value in this ecosystem lives on the appchains: Osmosis, Neutron, Stride, dYdX. None of it lives on the Hub that was supposed to be collecting rent from all of them. That's not a footnote to Part 3's argument about ATOM's monetization strategy. It's the data point.

What it means for the thesis

Three different security choices, all four chains still standing. Neutron bet on full sovereignty and left ICS after two years. Stride bet on borrowed security and, as far as the public record shows, is still renting it. Osmosis and dYdX skipped ICS entirely and ran their own validator sets from day one — one because it launched independent, the other because it walked away from a different chain's security model to build its own.

There's no clean headline here. ICS didn't fail across the board — it still works for Stride. It didn't succeed across the board either — it lost its flagship consumer chain. The honest read is that the model fits some chain profiles, capital custodians who need to borrow trust, and not others, fast-iterating DeFi hubs that need to control their own upgrade cadence.

The thesis's strongest evidence in this article is dYdX: a team with no Cosmos allegiance chose sovereignty anyway, and built something at real scale doing it. Its weakest evidence is the raw TVL table above — the appchain model produces real differentiation, but so far it hasn't produced size. Whether that changes depends partly on infrastructure still being built underneath all of this. Part 5 covers the next piece of that puzzle: Celestia and modular data availability — a bet that separating data availability from execution makes launching and securing the next generation of appchains cheaper than what Neutron, Stride, or dYdX had to do it the hard way.