PAXG’s design: a gold warehouse receipt on Ethereum

PAXG works because it embraces the boring part of “tokenized gold.” It is explicitly centralized, explicitly administered, and built to survive compliance reality. Paxos frames PAXG as a digital token backed by physical gold, where each token is backed by one fine troy ounce of gold stored in LBMA vaults in London, with the underlying metal held in custody by Paxos Trust Company under its gold backing model.

Legally and operationally, Paxos treats PAXG as ownership rights in allocated gold, not a governance or revenue-bearing crypto asset. The PAXG terms describe the token as “akin to a warehouse receipt” representing beneficial ownership of a pro rata portion of allocated gold.

Two details matter for tokenomics analysis. To map those levers cleanly, it helps to think in terms of core design components.

First, the token’s economic promise is redemption, not protocol growth. Paxos states PAXG can be redeemed for LBMA-accredited Good Delivery bars, redeemed for USD at current gold market prices, and (for institutional customers) redeemed for unallocated loco London gold.

Second, compliance controls are part of the product. Paxos contractually discloses the right to freeze and upgrade PAXG at an aggregate level. Those controls are also visible at the smart contract layer. The verified implementation includes pausing, address freezing, and “wipe” functionality for frozen addresses via an asset protection role.

Supply: elastic mint/burn with centralized controls

PAXG does not have “token emissions” in the way Web3 audiences usually mean it. There is no scheduled issuance, no staking inflation, no token sink meant to subsidize security. Supply is demand-driven. If more users want tokenized gold, new tokens are created. If users redeem, tokens are destroyed.

The control plane is explicit on-chain. PAXG’s verified contract includes a “supply controller” role and functions to increase and decrease supply. This is not a permissionless mint. It is an issuer-controlled mint/burn model.

That issuer control is reinforced off-chain. Paxos’ terms state only verified customers may purchase PAXG from Paxos or redeem/convert with Paxos, and Paxos may refuse issuance or redemption where it believes doing so would be contrary to law or would expose it to legal liability.

In other words: PAXG’s supply is “elastic,” but not “autonomous.” The elasticity is governed by Paxos’ onboarding, compliance, and operational constraints.

As a current snapshot (not a parameter), CoinGecko shows PAXG’s Ethereum contract address as 0x45804880de22913dafe09f4980848ece6ecbaf78 and lists a circulating supply of about 500,000 PAXG on March 5, 2026.

Fees and fiscal flows: where Paxos gets paid (and how holders can get diluted)

PAXG’s token economy is simple: holders do not earn platform revenue, and there is no native yield. The issuer monetizes through service fees around minting and redemption, plus potential future storage economics. For a second issuer-admin reference point, compare with PYUSD tokenomics.

Creation and destruction fees are the most direct fiscal flow. Paxos’ fee schedule states that when you buy or sell PAXG from your Paxos account (including conversions between PAXG and USD, gold bars, or unallocated gold), Paxos charges fees to process creation and destruction of tokens.

The current schedule is tiered and charged in PAXG. It includes (i) a minimum purchase size of 0.03 PAXG, (ii) a fixed 0.02 PAXG fee for orders between 0.03 and 2 PAXG, and (iii) percentage-based tiers for larger orders (1.000% from 2-25 PAXG down to 0.125% for 800+ PAXG).

There is also a time-bounded incentive that directly affects the minting economics. Paxos states that until March 31, 2026, PAXG creation fees are waived on purchases, after which the standard creation rates apply.

Paxos is careful about where those fees apply. The same fee page states these creation/destruction fees do not apply when purchasing and selling through Paxos’ exchange order book or elsewhere outside the Paxos wallet, though trading may be subject to exchange fees.

On-chain transfer fees are the other lever, and this is where history matters. Paxos’ current PAXG product page claims “zero on-chain transfer fees.” At the same time, the verified contract clearly retains a fee module: it has a configurable fee rate, fee recipient, and functions to set them (controlled by a fee controller role).

From a regulatory pragmatist lens, that combination is coherent. Turning transfer fees off reduces “issuer takes a toll on secondary market activity” optics. Keeping the capability allows Paxos to reintroduce on-chain monetization if it ever needs to cover custody economics without forcing users back onto its platform.

Storage fees are the sleeper risk and the most under-discussed tokenomics lever. Paxos’ help center states it does not charge gold storage fees to customers “at this time.” But the PAXG terms reserve the right to charge storage fees to all token holders by issuing new PAXG tokens to Paxos, explicitly describing this as dilutive to existing holders, with at least 30 days prior notice before implementation.

Finally, there is account-level fee surface area if you custody with Paxos. The PAXG terms include a $2 monthly charge (or equivalent in PAXG) if a Paxos account maintains a non-zero balance and has had no issuance or redemption activity for twelve months or longer.

Parameter control: “governance” is an admin keyset, not a DAO

There is no governance token. No DAO. No vote. That is a feature, not a gap, if your goal is to avoid blurry “utility plus profit expectation” narratives. For a governance-forward contrast, see DeXe tokenomics.

Instead, PAXG has a conventional issuer-admin model. On-chain, the verified contract exposes roles that can pause transfers, freeze addresses, wipe balances of frozen addresses, and control supply. Off-chain, Paxos states it has the ability and right to freeze and upgrade all PAXG tokens on an aggregate basis.

Upgradeability is also structural in the contract architecture. Etherscan shows PAXG is deployed behind an AdminUpgradeabilityProxy, which allows the implementation contract to be upgraded by the proxy admin.

This design choice carries a clean trade-off.

It lowers the probability of “governance as securities feature.” Holders do not participate in decision-making, and Paxos does not promise to route protocol cashflows back to the token.

It raises the probability of “issuer discretion as holder risk.” If you hold PAXG in a self-custodied wallet, you are still downstream of Paxos’ compliance posture and key management. Paxos’ terms state the freeze/seizure framework applies to all holders, even if they are not Paxos customers.

Market structure: pegs, redemption rails, and DeFi composability

PAXG’s price behavior is not enforced by an algorithm. It is enforced by arbitrage, and arbitrage depends on redemption rails.

On the Paxos platform, pricing for mint/redemption is tied to the London gold market. Paxos states the price of PAXG on paxos.com is based on streaming prices from StoneX representing the real-time London gold market, and that purchases and sales on paxos.com are only available during London gold market hours (with specific weekly closures described).

There is a second market regime: the order book. Paxos notes that PAXG/USD under the “Trade” tab is order-book based and can trade 24/7, while the “PAXG” tab quote is based on StoneX prices for PAXG minted or redeemed.

Redemption is also segmented. Paxos’ help center states that redeeming for a physical gold bar requires a minimum of 430 PAXG (plus fees), and that Paxos can securely deliver the bar only to vaults in the UK.

That 430 PAXG threshold matters for peg quality. It pushes most users toward “redeem to USD” or “trade out on exchanges,” not “pull bars.” Physical redemption exists, but it is a whale-grade backstop.

On transparency, Paxos offers two complementary proofs. It publishes monthly attestations for PAXG. It also runs a gold allocation lookup tool that lets on-chain holders look up serial number and bar information for the gold allocated to their Ethereum address.

Paxos’ help center describes allocation as automatic and continuous, including pro rata bar ownership when holders are below a full bar, with periodic reallocation to consolidate ownership.

Security and custody posture is disclosed, but it is not “trustless.” Paxos states its vault provider maintains insurance against loss of the gold held in custody for token holders, including in vault facilities and in transit. It also states it publishes monthly attestations, and its transparency page notes that attestations posted on or after February 28, 2025 are issued by KPMG LLP (with earlier attestations issued by Withum).

DeFi composability is the double-edged sword. PAXG can be used as collateral or liquidity elsewhere because it is an ERC-20 token. That composability can create “yield on gold,” but the yield is always external and counterparty-dependent. PAXG itself does not distribute fees or protocol income to token holders.

Risk analysis

PAXG’s core proposition is legally conservative tokenization: regulated issuer, allocated backing, audits, and explicit enforcement controls. The design is coherent.

It also concentrates risk. Not in the usual “runaway inflation” way. In issuer discretion, enforcement hooks, and the operational reality that redemption is permissioned.

Dominant risk: issuer enforcement and discretionary control over transferability and redemption.

Paxos’ terms reserve broad powers: the ability to freeze and upgrade all PAXG tokens on an aggregate basis, compliance with legal directives, and (in certain circumstances) refusal of redemption. The same terms state that if Paxos determines after investigation that PAXG has been used for illegal or sanctioned activity, it may not permit redemption, and the PAXG and allocated gold backing it may be forfeited.

Those are not abstract legal threats. They map directly onto the smart contract surface area. The verified implementation includes an asset protection role that can freeze addresses and wipe frozen addresses. It also includes pausing controls.

From a compliance perspective, that is the point. If you want institutional distribution, you need the ability to enforce sanctions, respond to court orders, and remediate compromised flows. PAXG is designed to be compatible with that world.

From a holder perspective, this is the axis that dominates every other risk. A user can self-custody PAXG, but they cannot self-guarantee transferability, censorship-resistance, or redemption access. Even if you never touch Paxos’ platform, your token is still governed by a contract that embeds these controls, and by terms that explicitly extend enforcement actions to all holders.

That does not make PAXG “bad.” It makes it legible. If your product, treasury policy, or DeFi vault assumes censorship resistance, PAXG is the wrong primitive. If your product assumes regulated redemption and you can tolerate issuer discretion, PAXG is structurally aligned.

There is one more nuance: upgradeability. Etherscan shows PAXG is deployed behind an upgradeable proxy. Paxos’ terms also mention upgrade rights. That means key economics can change without a token-holder vote. Today’s “zero on-chain transfer fees” claim, for example, coexists with a contract that retains fee controls.

That is a compliance-friendly design. It is not parameter-stable in the way DeFi-native assets try to be.

Top 3 risks

  1. Issuer enforcement (freeze, wipe, redemption denial). Trigger: regulatory action, sanctions screening hits, suspected illicit flows, or Paxos’ own discretion under its terms. Mechanism: asset protection role can freeze addresses and wipe frozen balances; Paxos can refuse issuance/redemption and can freeze access to PAXG and backing gold. Who bears it: self-custody holders, DeFi protocols accepting PAXG collateral, and any downstream counterparties. Indicators: on-chain freeze/wipe events, proxy upgrades, public enforcement disclosures, and widening on-exchange spreads during enforcement news.

  2. Economic drift via fees (including potential storage-fee dilution). Trigger: Paxos changes fee schedules, re-enables on-chain transfer fees, or introduces storage fees. Mechanism: (i) platform-level creation/destruction fees directly tax mint/redeem; (ii) terms allow storage fees implemented via issuance of new PAXG to Paxos, diluting holders. Who bears it: long-term holders, and any strategy whose expected returns are thin (basis trades, low-margin lending). Indicators: updates to the Paxos fee schedule, new storage-fee notices, and on-chain changes to fee parameters via the fee controller role.

  3. Peg and liquidity dislocations around market hours and redemption constraints. Trigger: London market closures, stressed crypto liquidity, or redemption friction (notably the 430 PAXG physical redemption minimum and UK-vault delivery constraint). Mechanism: mint/redeem is limited by platform hours and KYC rails, while secondary markets can trade 24/7, allowing basis gaps to open. Who bears it: traders relying on tight tracking, DeFi LPs, and protocols marking collateral to oracle prices during stress. Indicators: persistent premium/discount versus spot gold proxies, deteriorating order book depth, and increased redemption-related support activity.

If you are integrating PAXG into a product, treat the issuer controls as first-class inputs to your token economy design. That is where most “surprise risk” lives, not in emissions math. If you want a reusable evaluation checklist, our tokenomics methodology explains the approach behind these reviews.

For teams doing tokenomics consulting around collateral standards and RWAs, PAXG is a useful case study precisely because it makes the compliance trade-offs explicit. If you need support designing around those constraints, our tokenomics services cover collateral standards and RWA integrations.



This article is part of our Tokenomics Deep Dive series.