AXL exists to secure the router, not to “bridge” for free

Axelar’s core bet is that cross-chain routing should be secured like a chain, not like a product. That means a delegated proof-of-stake validator set, threshold cryptography, and on-chain governance controlling what gets connected and how messages get finalized. The AXL token is the stake, the vote, and the fee unit that makes that machine run.

The security model is explicit about decentralization and explicit about thresholds. The Axelar whitepaper states a 90% safety threshold, meaning “almost all validators” must collude to withdraw locked funds or forge state proofs. That is a hard-line choice. It pushes compromise resistance up, while making liveness more fragile under validator outages. Axelar acknowledges that and pairs it with fallback recovery mechanisms, including an “emergency unlock key” on connected chains and a “secondary recovery set” intended to be maximally open if the primary network stalls.

From a decentralization purist lens, this is the right axis to argue about. Not “did they use MPC.” Most bridges do. The real question is: who can change the validator set, who can change thresholds, and what stake distribution makes those promises meaningful. For a useful comparison, see our Wormhole tokenomics review.

Supply, genesis distribution, and the uncomfortable truth: no hard cap

AXL started with 1,000,000,000 tokens created at the genesis block. The initial allocations are described consistently in public disclosures.

Those categories were “subject to their own release schedules,” and the “original release schedule” is described as starting on September 27, 2022. Public docs do not, in one canonical place, provide a clean vesting table for every bucket. That matters for modelability and for governance realism, because “widely distributed” is not a vibe. It is an unlock function. If you want a structured checklist for modeling these levers, our design components guide is a useful baseline.

Now the part many people try to talk around: AXL does not have a fixed maximum supply. The same disclosure set states that additional AXL may be issued over time. CoinGecko currently shows max supply: ∞.

As of CoinGecko’s on-chain referenced figures, AXL shows Circulating Supply: 1,130,325,757 and Total Supply: 1,233,252,698. That alone tells you the system has already moved beyond the genesis issuance in net terms. Whether that expansion continues is governance-dependent, and that is the central tokenomics risk surface.

Issuance, burns, and where value flows now

Axelar’s tokenomics pivot since early launch has been about removing “connect more chains = mint more tokens” as the default growth policy. That culminated in the v1.2.1 upgrade, code-named “Cobalt”, which Axelar says was completed on February 25, 2025 following tokenholder approval via on-chain governance in the Cobalt upgrade.

Cobalt changes fee routing in a way that is mechanically legible. Axelar states that 98% of AXL gas fees are sent to a burn address, while 2% still goes to the community proposals grant pool. This is a clean fiscal flow. It says: network usage reduces supply, while the on-chain grants pool stays funded.

The same post describes how fee abstraction works. Users pay “once in the source-chain token,” and Axelar Gas Services converts that payment into AXL and any other gas tokens needed to complete the cross-chain transaction. So AXL demand is not only “people decide to pay in AXL.” It can also be mediated by protocol services that buy AXL as part of execution.

On the issuance side, current minting parameters show standard Cosmos minting inflation set to 0%, with “annual provisions” also effectively 0. That does not prove “no new issuance is possible.” It proves the current mint module is not providing issuance. In an uncapped-supply chain, that is still a governance decision, not a constitutional limit.

Axelar’s growth incentive has shifted toward explicit, per-connection reward pools. Cobalt describes that “new blockchains connect by setting up reward pools out of the existing AXL supply” to reward Verifiers that support each new connection.

Two implications matter.

First, chain growth can become less dilutive if reward pools are funded from existing supply and/or by integrators, rather than by protocol-level inflation. That is directionally good for holders. It is also a centralization pressure point because it can privilege chains and partners that can afford to seed pools at meaningful size.

Second, Axelar is leaning into “explicit budgets” for security work. That is healthier than pretending consensus is free. But it increases governance load. Someone decides reward rates, eligibility, and verifier requirements. If those decisions are not credibly decentralized, you just moved centralization from the issuance curve to a budget committee.

Staking design: decentralization knobs and their limits

Axelar’s validator set is capped at 75 active validators. The unbonding period is 7 days. Slashing parameters include 2% for double-signing and 0.01% for downtime.

In isolation, 75 validators is not terrible for a cross-chain system. It is still a small council compared to “global” security systems, and Axelar’s own whitepaper argues threshold cryptography allows the validator count to be “as large as possible.” The chain chooses a cap anyway, which is an operational reality. More validators means more coordination burden, more heterogeneous infra, and more ways to fail in cross-chain signing.

Axelar’s Foundation explicitly frames delegation policy as promoting decentralization. It describes validator operation as “permissionless,” while noting you need “sufficient stake” to enter the active set. That is standard for Cosmos-style PoS. The decentralization question is not “can anyone run a node.” It is “can new validators realistically attract stake away from incumbents, and can delegates coordinate to avoid oligopoly.”

Here’s a blunt data point: one chain analytics view reports that the top-100 staking accounts control 96.73% of voting power. Even allowing for methodology quirks, that number is a red flag for governance decentralization. In a system where governance is stake-weighted, concentrated stake means concentrated control, even if the validator list looks diverse on paper.

Axelar also describes limits on voting power concentration via quadratic voting for cross-chain operations. That is interesting. It suggests Axelar distinguishes between (1) chain governance and (2) cross-chain verification/signing weight, and tries to keep “more stake” from translating linearly into “more control” in at least one of those domains. The problem is that the most critical control plane still includes governance itself, which can change parameters, upgrade code, and define what “cross-chain operations” even mean.

Finally, remember the earlier 90% threshold. If a large minority of validators go offline, the protocol’s cross-chain signing can stall, even if the chain keeps producing blocks. Axelar explicitly anticipates this and builds fallback recovery. From a purist perspective, fallback mechanisms are where decentralization goes to die. You do not want to discover in a crisis that “emergency unlock” is practically controlled by a small group, even if it is theoretically broad.

Governance: stake-weighted control with low friction and high blast radius

Axelar governance looks like a standard Cosmos-family on-chain system. Current parameters include:

Minimum deposit: 2,000 AXL, voting period: 3 days, quorum: 33.4%, threshold: 50%, and veto threshold: 33.4%.

Those numbers are not cosmetic. They define the governance capture surface. A 33.4% quorum means a proposal can pass with participation just above one-third of staked voting power. If stake is concentrated and voter apathy is chronic, “one-third shows up” can become “a small cluster decides.” The 33.4% veto threshold gives a blocking minority, but only among those who vote.

Axelar is also explicit that connectivity is governance-mediated. The Cobalt post states that “new connections to the Axelar Network are subject only to on-chain votes by AXL tokenholders.” That is good on paper. It means no closed committee can silently add a chain. In practice, it raises the bar for decentralization of the voter base because each new connection can change the economic and security posture of the whole network.

Governance has already made structural changes. For example, a parameter change proposal to increase the active validator set to 75 is recorded as Proposal #164, with the description stating it “increases validator active set to 75.” Earlier governance forum discussion explicitly connected validator set sizing with inflation settings and incentives for external chain verification.

This is the part progressive decentralization narratives gloss over. If parameter agility is a feature, then governance decentralization is the product. With AXL, the chain has an uncapped supply design, protocol-level fee burning, and per-connection reward pools. All of these can be modified by governance and upgrades. The token is not just an asset. It is the control surface for a cross-chain security perimeter.

Risk analysis

The attractive story is simple: AXL secures cross-chain messaging, fees are burned, and chain growth no longer demands inflation. The structural reality is harscher. Axelar’s security guarantees depend on (1) validator and stake decentralization, and (2) governance that cannot be quietly captured. Axelar can be “more decentralized than other bridges” and still be structurally centralized in the ways that matter during a crisis.

Top 3 risks

  1. Governance capture through stake concentration (dominant): Trigger: sustained low governance participation near the 33.4% quorum combined with increasing stake concentration among a small set of entities. Mechanism: stake-weighted voting can approve upgrades and parameter changes, including economic policy (minting settings), validator set rules, and connection governance, even when the broader holder base is inactive. Who bears it: AXL stakers (dilution or adverse parameter shifts), integrators (policy uncertainty), and users whose cross-chain assets rely on Axelar’s validator honesty. Measurable indicators: (i) share of voting power controlled by top staking accounts (for example, the reported 96.73% for top-100 staking accounts), (ii) repeated proposals passing with turnout just above quorum, (iii) recurring upgrades that materially change monetary policy (minting on/off) or fee routing.

  2. Liveness failures at high safety thresholds: Trigger: correlated validator outages, censorship events, or operational shocks that take a meaningful minority of validators offline. Mechanism: Axelar’s design targets a 90% safety threshold for cross-chain security, which can increase the chance of stalls if the network cannot gather enough signing participation. The system then leans on fallback recovery, including emergency unlock keys and a secondary recovery set. Who bears it: users with time-sensitive cross-chain flows, protocols depending on message finality, and tokenholders if confidence shocks force governance into “emergency mode.” Measurable indicators: rising validator downtime and jailing events (given defined slashing rules), delayed cross-chain finality, and increased discussion or activation of emergency recovery processes.

  3. Monetary-policy unpredictability in an uncapped asset: Trigger: macro downturns, validator profitability stress, or incentive shortfalls for verification work that motivate governance to re-enable or expand issuance. Mechanism: AXL is explicitly described as having no fixed supply and the ability to issue additional tokens over time, while current on-chain minting inflation is set to 0%, which is a policy setting, not a guarantee. Who bears it: long-term holders and stakers if issuance increases faster than demand from fees, burns, and reward-pool funding. Measurable indicators: governance proposals targeting minting and incentives, changes in on-chain mint parameters away from 0%, and accelerating divergence between circulating supply and net burn effects.

Dominant risk: governance capture through stake concentration

The dominant risk is not “bridge hacks” in the abstract. It is that Axelar’s most important decentralization guarantees live above the token layer, but are ultimately controlled by it.

Start with the governance mechanics. AXL governance has a 33.4% quorum and 50% pass threshold. That is a common Cosmos configuration. It is also a configuration that can produce “legitimate” outcomes from a minority of total staked power when turnout is weak. In a chain whose job is to be the security perimeter for cross-chain value, that is a large blast radius for low participation.

Now pair it with concentration. A reported 96.73% of voting power controlled by top-100 staking accounts is an extreme number. Even if you discount it, the direction is clear. Stake is not evenly distributed across a huge base of independent voters. So the real governance system becomes: a small cluster of large delegators and large validators coordinate (or fail to coordinate), and everyone else rationally free-rides.

That coordination problem would be “just a token politics issue” if Axelar were a normal chain. It is not. Axelar’s whitepaper explicitly ties security to validator decentralization and threshold assumptions. If the stake-weighted control plane centralizes, you can end up with a network that is technically threshold-secure, yet socially and economically brittle. A small set of actors can influence upgrades, connection policy, and incentive budgets, then credibly claim it was “decentralized” because an on-chain vote happened.

Fee burning does not fix this. Burning 98% of AXL gas fees is a monetary policy choice, and a well-defined one. But a burn is not decentralization. If governance is capturable, the burn can be changed, bypassed, or offset by issuance. The system is uncapped by design, and current minting being 0% is a parameter state, not a constitutional constraint. If you’re building monitoring around these risks, our research reports can help you operationalize what to track.

Reward pools and verifier bonds help align incentives at the edge. They also introduce more governable surfaces: pool parameters, verifier requirements, chain onboarding policies. Those surfaces only remain credibly neutral if the governance base remains credibly decentralized.

If you are integrating Axelar into a protocol where cross-chain execution is existential, treat governance decentralization as part of your threat model. Monitor turnout, concentration, and upgrade cadence as diligently as you monitor audits and validator uptime. If you need tokenomics consulting on verifier incentives or token economy design around reward pools and fee policy, do it with governance constraints front and center, not as an afterthought.



This article is part of our Tokenomics Deep Dive series.