Earlier this year, Cosmos Labs kicked off a multi-phase research process to redesign ATOM’s economic model to move ATOM toward sustainability. Following a competitive RFP process, Gauntlet was selected as the quantitative research partner for this work.
Today we’re sharing the results of Phase 1: a comprehensive, empirical analysis of ATOM sell pressure, holder behavior, and staking economics.
Full Phase 1 Report: Gauntlet: ATOM Inflation and Sell Pressure Analysis
Bottom line
Phase 1 gives us the empirical foundation we needed, and it points to a more precise problem than “inflation is too high.” Gauntlet’s on-chain analysis shows that ATOM sell pressure is small relative to supply and did not accelerate, that it routes almost entirely through a handful of large holders into centralized exchanges, and that the single largest sell reaction on record came from a governance decision, not a market crash. The strategic read is that staking demand is real and sticky (staking TVL at record highs), and the community’s sensitivity to token-economic changes is itself evidence that inflation is a demand driver, not merely a cost.
To be clear: This doesn’t mean that inflation is 100% blameless. The Hub still releases the largest immediately liquid reward footprint of any major PoS chain, roughly 3.6x NEAR and 5.7x Ethereum. The right question for Phase 2 is not “how much inflation?” but “what can stakers offer the network in return for yield, and how do we better manage immediately liquid issuance without triggering the kind of repositioning Prop 848 caused?”
What Phase 1 established
Reward-driven selling is small, but Cosmos exposes more liquid reward than any peer. Of 8.18M ATOM in rewards withdrawn over the January to March window, 42.6% reached a sell-like route in the same week, 27.7% was re-staked, and 29.7% stayed liquid but unsold. Same-week reward selling averaged only about 0.063% of supply per week and never trended higher. However, Cosmos Hub releases 0.153% of supply per week in claimed rewards, roughly 3.6x NEAR and 5.7x Ethereum. We inflate meaningfully more than our peers even if those rewards are not the dominant driver of daily selling. That finding rules out reward selling as a main driver of price, but it doesn’t rule out how issuance impacts dilution.
Sell pressure concentrates in a few large holders, routed to CEXes. Across every study, 95 to 98% of sell-like flow went directly to centralized-exchange deposit addresses, with Coinbase the dominant destination; DEX swaps were negligible. Roughly 44 wallets drove the majority of sell pressure over the last six months, and whales plus mega-whales accounted for about 79% of event-window selling and 51% of cohort flow. This is a large-holder, single-channel phenomenon, not broad retail panic. Net-flow analysis confirms the cohort is a net distributor (about 7.56M ATOM net, roughly 28% of gross outflow), concentrated in a handful of addresses.
One important temper on the CEX figure: a meaningful share of exchange-bound flow is not holders selling. The largest single exchange route (an Upbit validator address, about 25% of reward-route volume) is almost certainly the exchange claiming and managing its own validator commission, not customers depositing ATOM to sell. Gauntlet recommends that ongoing monitoring report a retail-adjusted view that segments out exchange-operated validator and staking addresses, so the headline gross figure is not read as purely discretionary retail selling. This adjustment changes who is selling, not how much is being issued.
Governance, not market stress, is the biggest sell catalyst. Prop 848 passing produced a seven-day reaction of 10.25% of supply, 21x the cohort’s baseline selling. That single governance outcome moved more supply than the FTX (2.49%) and Terra/LUNA (1.18%) collapses combined. Holders also reacted to finality, not discussion: the Prop 848 forum post drew only 0.62% of cohort outflow five weeks before the vote spiked to 10.25%. Large holders wait for a binding outcome before repositioning. This underscores why broad stakeholder alignment ahead of any on-chain proposal is non-negotiable.
How we are reading it
-
Staking demand is real and the community’s reaction proves it. Record staking TVL and the intensity of the Prop 848 response both show that ATOM holders care deeply about the token’s economic thesis. Sensitivity is a signal of engagement, not just risk.
-
We still have a structural reason to act. ATOM’s liquid-reward footprint is the largest among comparable networks. Optimization is justified on structure alone, independent of how much of today’s selling is reward-sourced.
-
CEX sell pressure is the key Phase 2 focus, and we frame it as a demand opportunity. The concentration of flow through a small set of exchange venues is not only a leak to manage. It is also an opening to broaden where and how ATOM staking demand is sourced (for example, expanding access through additional regulated venues to dilute any single exchange’s share), turning a sell-side concentration into a more distributed, resilient demand base.
-
The core design question is what stakers give back for yield. Rather than paying yield for passive lockup, Phase 2 explores what stakers can provide in return: liquidity provision, attestation, intent liquidity, and other productive contributions to the network. Staking-mechanic reform, not a blunt inflation cut, is the real lever.
What this means for Phase 2
Gauntlet’s proposed path is to optimize realized liquidity rather than headline emissions, and to tie any reduction to observed demand and sell-pressure conditions rather than a one-off binding cut. Some of the areas we plan to explore first as part of phase 2:
-
Dynamic, feedback-driven issuance. A PID-controller-style mechanism (modeled on NEAR’s approach) that steps emissions down in small, continuous increments while whale and CEX-bound pressure stays at or below baseline, and eases when abnormal pressure appears. The natural trigger to cut inflation is when product demand is strong enough that dilution kicks in organically.
-
Smooth or delay reward liquidity. Streaming rewards, delayed claim windows, stronger auto-compounding defaults, and longer-duration staking incentives, all evaluated against Claimed/Supply and Sold/Supply rather than headline inflation.
-
A Security Floor framework. An Economic Security Ratio dashboard (cost-to-attack versus benefit-to-attack, as Gauntlet built for NEAR) so we can reduce inflation as sell pressure allows while keeping a continuous read on validator-set security.
-
Governance market-risk discipline. Every major token-economic proposal should carry a market-risk appendix with expected abnormal-pressure ranges, whale and CEX monitoring, and staged communications timed separately for discussion and vote-finality.
The tension Phase 1 hands to Phase 2 is clear: our structural liquid-reward footprint is high and should come down, yet the largest sell reaction in ATOM’s history was the market’s response to exactly that kind of binding change. The Phase 2 mandate is to reduce the footprint without triggering that repositioning.
Phase 2 scoping is underway now. Consistent with our plans, community and validator input will come throughout the research before any idea reaches an on-chain vote, with discussion and vote-finality communications staged separately.
What this means for the Hub roadmap
This feeds the Hub roadmap directly. Previously there was a tension: we knew inflation was attractive, but we did not know whether the Hub could retain or coexist with that model given the inferred impact of issuance. Phase 1 resolves much of that tension. It confirms that the Hub holds an asset most networks do not: a large, sticky stake base whose inflation is not a major driver of sell pressure. Staking TVL sits at record highs and held through everything short of a binding governance change, which tells us yield has a real role to play. The question this creates for the roadmap is a more productive one: what can the inflation model support?
Today, staked ATOM does a single job, securing the network, and earns a single reward, issuance. The roadmap changes that by making stake a working input to Hub products: stake that provides liquidity, attests, backs intent execution, or secures new services. Each of those is a service someone pays for, and because the stake itself delivers the service rather than only the protocol around it, stakers could hold a direct claim on that revenue. More uses for stake mean more fee streams, and those fee streams do not have to be burned against inflation to matter. They can flow to stakers as rewards that exist because the stake did something, not because the protocol issued more supply.
This is what makes adjusting inflation, or immediately liquid issuance, viable without prompting holders to exit. The point is not that product revenue must match inflation line for line, but that the reward base diversifies: security yield becomes one income stream among several rather than the only reason to stay. Phase 1 found that inflation currently acts as a demand driver, so the roadmap’s job is to evaluate how best to optimize its impact, so we can continue to leverage it as a resource, build demand drivers stronger than issuance, and pay stakers from them. We will carry this into the Hub roadmap process as a design input.