ATOM, Interchain Security, and the Value-Accrual Problem

What ATOM actually does

ATOM secures the Cosmos Hub through delegated proof-of-stake. Validators bond ATOM as collateral, run the nodes that produce and finalize blocks, and get slashed — a portion of their bonded ATOM destroyed — for downtime or for double-signing (broadcasting two conflicting blocks at the same height). Token holders who don't run a validator can delegate their ATOM to one and share in the rewards and the slashing risk. ATOM is also the Hub's governance token: voting power is proportional to bonded ATOM, so the people with the most capital at stake decide how the chain's rules change.

That's the part that works exactly as designed. The part that didn't is the pitch Part 1 covered in detail: ATOM as the base-layer gas and reserve asset for a universe of sovereign chains connected by IBC. In that version of the story, every appchain that plugged into the Cosmos network would end up routing some economic activity — fees, liquidity, settlement — back through ATOM, the way ETH captures a cut of nearly everything that happens on Ethereum.

That never materialized. When a user moves an asset across chains using IBC, the transaction fee is paid in whatever token the executing chain charges gas in — Osmosis fees are paid in OSMO-denominated gas, Injective fees in INJ, and so on. IBC is a message-passing protocol, not a toll road with the Hub sitting at the crossing. The Hub doesn't get a cut of a swap on Osmosis or a trade on dYdX just because IBC carried the packet. ATOM captures essentially none of the fee volume moving across the ecosystem it helped connect.

That leaves the Hub's own transaction fee revenue — small, because the Hub is mostly a settlement and governance chain, not a high-throughput application platform — set against ATOM's staking rewards, funded almost entirely by inflation (new ATOM minted and distributed to stakers, not revenue collected from users). Day to day, ATOM's utility is overwhelmingly about ATOM: stake it, vote with it, get paid in more of it. It is not, in practice, a claim on external economic activity happening elsewhere in the ecosystem. The rest of this article describes what Cosmos governance has tried, and is still trying, to do about that.

The inflation wars: ATOM 2.0 to Prop 848

In 2022, a whitepaper called "ATOM 2.0" proposed a sweeping overhaul of ATOM's tokenomics — new vaults, new issuance curves, new revenue mechanisms. Cosmos governance rejected it. It was never implemented. It's worth naming only because it still circulates online as if it were a live roadmap; it isn't. It's a dead reference point.

The tokenomics change that actually happened came a year later. Prop 848, the "ATOM Halving," was approved on 25 November 2023 with the highest voter turnout in Hub governance history. It cut the chain's maximum inflation rate from 20% to 10%.

Two months later, a follow-up went further and failed. Prop 868, voted 23 January 2024, would have cut the inflation floor from 7% down to 0%, but only once the bonded ratio (the share of circulating ATOM actively staked) passed two-thirds. It was rejected — 48.6% against versus 25% for — with several large validators arguing that pushing inflation toward zero once enough ATOM was staked would undercut the economic security of the chain, since lower rewards could push validators and delegators to unbond, lowering the very ratio the proposal was conditioned on.

As of July 2026, the Hub's live on-chain mint parameters are: inflation_min = 7%, inflation_max = 10%, goal_bonded = 67%. Current inflation is sitting at the 10% ceiling — which means the bonded ratio is still below the 67% target. Inflation isn't falling toward the floor; it's pinned at the post-848 maximum.

A later governance action, Prop 996 in 2025, changed where that inflation goes rather than how much of it exists: the staker share of new issuance rose from 90% to 98%, while the community pool's share fell from 10% to 2%. That's a redistribution of the existing pie between two groups who both already held ATOM — not a reduction in how much new ATOM enters circulation.

Put simply: ATOM is still meaningfully inflationary, and there is no live mechanism that burns supply or otherwise offsets that issuance with real fee revenue.

Interchain Security v1: renting out the validator set

If ATOM couldn't tax the ecosystem's transaction volume, the next idea was to rent out its validator set directly. Interchain Security — specifically its first version, called Replicated Security — lets a new chain, called a consumer chain, use the Cosmos Hub's entire existing validator set and its bonded ATOM as its security, instead of bootstrapping its own set of validators and its own token to stake. The Hub, in this arrangement, is the provider chain: it supplies the validators and the capital at risk. In exchange, the consumer chain shares a cut of its own transaction fees and MEV (value extracted from transaction ordering) with the Hub's ATOM stakers.

Only two chains ever adopted full Replicated Security. Neutron, a smart-contract platform, became the first Cosmos consumer chain in 2023, sharing roughly 25% of its transaction fees and MEV with the Hub. Stride, a liquid-staking protocol, followed via Prop 799 with 97% approval, sharing 15% of its revenue.

The pitch at launch was that this would be the default path for new Cosmos appchains: rather than each one recruiting and paying its own validator set from scratch, dozens would queue up to rent the Hub's already-established security. Adoption stalled at two.

The numbers: what ICS actually paid ATOM stakers

This is where the mechanism's performance can be measured directly, and the numbers are unambiguous.

The full-year 2024 combined figure roughly tripled what the mechanism produced the year before. That's real growth. It was also, in the same year, less than 0.12% of what the Hub spent minting new ATOM to pay its stakers.

Separate from the revenue side, independent validator-economics research found that around 40% of active Hub validators operate at a loss once the infrastructure cost of running consumer chains is counted — extra servers, extra monitoring, extra operational risk for running Neutron or Stride on top of the Hub. The same research found that a single slashing event triggered by a fault on a consumer chain could wipe out centuries' worth of accumulated ICS revenue for a mid-sized validator in one incident.

The mechanism was built specifically to let ATOM's staked capital earn money from the ecosystem it secures. It earned ATOM stakers a rounding error.

Neutron's exit: the Mercury upgrade

In April 2025, Neutron — the first chain to adopt Replicated Security back in 2023, and by then the longest-running Interchain Security consumer chain in Cosmos — migrated off ICS entirely. Its "Mercury" upgrade stood up a fully independent validator set and introduced native NTRN staking, replacing the Hub-supplied validators it had relied on since launch.

Neutron's own stated rationale is worth quoting rather than summarizing, because it doubles as the clearest verdict on the whole model: ICS, in its assessment, "didn't create significant value accrual for ATOM," "placed a heavy burden on the Cosmos Hub's validators," and "slowed down the upgrade process for chains using ICS." (Neutron's broader trajectory as an independent chain is covered in Part 4.)

What matters here is what the exit signals: the flagship proof-of-concept for renting ATOM's security chose full sovereignty the moment it had the option. As of mid-2026, the roster of full Replicated Security consumer chains has effectively shrunk to Stride alone. No major new chain has stepped in to replace Neutron under the original model.

Partial Set Security and the AEZ retry

Cosmos governance's response to the adoption problem was Partial Set Security, or ICS v2 — a redesign meant to fix Replicated Security's core flaw. Under the original model, every single Hub validator was forced to validate every consumer chain, whether or not they wanted the work or found the compensation worthwhile. PSS let each consumer chain instead choose "Top-N" (a required slice of the Hub's largest validators by stake) or a fully voluntary "Opt-in" model.

The redesign cleared its CHIP (Cosmos Hub Improvement Proposal) signaling vote with 99.6% approval in February 2024, and shipped to the Hub in the Gaia v17 release.

Alongside PSS, the Hub has promoted the "ATOM Economic Zone" (AEZ) — a branding umbrella positioning Stride and (until its exit) Neutron as founding members of an ATOM-aligned ecosystem, partly coordinated through DoraHacks quadratic-funding grant rounds. Round 3, in roughly April 2025, distributed about 8,000 ATOM plus about 31,111 USDC across 14 projects.

Honestly assessed, the AEZ functions in practice more as a grants and marketing umbrella than as a proven value-capture mechanism. The chains most associated with it either generate negligible ICS revenue or have already left the security-rental model behind.

The chains that never asked

While ATOM's stewards were redesigning the rental model, several of Cosmos's most successful chains simply skipped it.

dYdX Chain (v4) launched on mainnet on 26 October 2023 as a fully sovereign Cosmos SDK layer 1, with its own validator set — more than 60 validators at launch. It used Cosmos tooling. It never rented Hub security. Celestia launched its mainnet and TIA token on 31 October 2023 running its own independent proof-of-stake validator set — 100 validators at launch — and was never an ICS consumer chain either. Osmosis, Injective, and Cronos have each operated for years as independent Cosmos SDK chains with their own tokens and their own validator sets, owing the Hub no fees and no governance deference.

The market tells you where the capital ended up.

Every chain in that table except ATOM itself bypassed the ICS/AEZ thesis entirely, and four of the five have a larger market cap than ATOM. ATOM itself trades around $1.54–$1.59 as of July 2026 — roughly 96% below its September 2021 all-time high of $44.77, a decline that predates and has outlasted every initiative built to reverse it.

2026's reckoning: COSMOSIS and the tokenomics rewrite

By late 2025, this record had become hard to argue with from inside the ecosystem too. Cosmos Labs, engaged by the Interchain Foundation, opened a formal, multi-stage research process with an explicit goal: replace ATOM's inflation-dependent staking model with a "non-circular," fee-driven system instead. Nine research proposals were submitted by a January 2026 deadline; selection and analysis were still underway as of mid-2026.

Then, on 11 March 2026, Osmosis put forward a far more drastic idea. Governance proposal "COSMOSIS" (#1029) proposed converting all circulating OSMO into ATOM at a fixed rate of 1.998 OSMO to 0.0355 ATOM, and folding Osmosis's AMM and liquidity infrastructure directly into the Cosmos Hub — effectively merging Cosmos's largest DEX into ATOM itself.

It failed, narrowly, in April 2026: about 50.37% voted no against 39.99% yes, with 9.61% abstaining. The debate didn't end there. After OSMO surged roughly 185% in 24 hours on 11 May 2026, discussion of a revised, non-dilutive version of the merger reignited, and remains unresolved.

Read together, a near-miss vote to absorb an entire sister ecosystem's application layer into the Hub, alongside a ground-up research mandate to replace how ATOM is issued in the first place, are not the marks of a value-accrual model that's working. They're the Hub publicly reaching for structural fixes because the organic thesis — appchains routing value back to ATOM on their own — didn't deliver.

The honest verdict — and the restaking parallel

Interchain Security and Ethereum restaking are the same basic business model, arrived at independently: take capital that's already staked securing one chain, and rent its economic security to a new protocol for a fee. Comparing the two at scale is instructive. EigenLayer, Ethereum's dominant restaking protocol, crossed $18 billion in restaked ETH across roughly 1,900 active operators by February 2026 — about 94% of the entire restaking market. Cosmos ICS, pursuing the identical idea for over three years, attracted two chains total, and generated ATOM stakers about $426,000 in a full year against an inflation bill north of $300 million annually.

The gap isn't explained by Cosmos's technology being worse. It isn't. IBC works as advertised, the Cosmos SDK is widely used, and several SDK-built chains — Celestia, dYdX, Injective, Cronos among them — are some of the more successful independent layer 1s in the industry. The gap is that none of that success flows back to ATOM by default, and every purpose-built mechanism designed to route it back — Replicated Security, Partial Set Security, the AEZ — has, by the ecosystem's own published numbers, failed to close it.

Cosmos the technology stack succeeded. ATOM the value-accrual thesis has not, and as of mid-2026 Cosmos governance itself is treating that as settled enough to justify rewriting ATOM's tokenomics from the ground up. Part 5 turns to Celestia — a chain built with Cosmos tooling, running its own validator set, that owes the Hub nothing and never asked to. The irony is hard to miss once you've seen the numbers behind it.