I keep a spreadsheet that nobody in my newsroom finds interesting. Last week Token Terminal pushed out a set of figures โ RWA on-chain at $46.4 billion, tokenized gold at $5.1 billion, gold claiming 11% of the category โ and I did what I always do. I added the pieces up by hand.
XAUT: $2.7 billion. PAXG: $1.9 billion. KAU: $231 million. PGOLD: $85.1 million. XAUM: $66.1 million. Call it $4.98 billion, the remainder being long-tail dust. It reconciles with the headline. It also tells you something the headline buried: two names hold roughly 90% of the entire tokenized gold float.
Headlines count. Ledgers concentrate. That gap between the announcement and the arithmetic is where the real story lives, and it is the thread I want to pull for the next thousand words.
Context: what a gold token actually is
The correction I had to make in my own notes was structural. Tokenized gold is not a project token. There is no team allocation, no vesting cliff, no unlock calendar, no staking APR, no governance vote. XAUT and PAXG both went live in 2019 as ERC-20 wrappers around metal sitting in a vault โ warehouse receipts with a wallet address. If your analytical framework assumes insider supply and emission schedules, you are modelling the wrong object entirely.
That matters because it locates the risk. There is no smart contract exoticism here; the contracts are standard templates. The engineering is not cryptographic. The engineering is trust โ audit frequency, redemption KYC, and whether the issuer publishes verifiable proof of reserves at all. XAUT has historically lagged on real-time on-chain attestation. PAXG, issued through Paxos Trust Company under NYDFS supervision, publishes periodic attestations and carries a regulatory moat the others simply do not.
The second correction concerns the denominator. A $46.4 billion RWA figure almost certainly excludes stablecoins โ include them and you are immediately north of $150 billion and the whole taxonomy collapses. That definitional boundary is not a footnote. It is the difference between "RWA is a $46 billion market" and "RWA is a rounding error next to the thing everyone actually uses."
There is a discipline to this that I learned the hard way. In 2017 I was running three parallel threads on Proof-of-Stake mechanics and growing an audience faster than I was verifying claims โ enthusiasm outrunning rigor, and it worked, which was the problem. The Terra unwind five years later installed the correction permanently: read the definition before you read the number.
Core: the number that isn't a number
Here is the thing I could not stop circling.
Tokenized gold's market cap is a product of two inputs: the number of tokens outstanding, and the price of gold. Nobody in the reporting distinguished between them. If gold ran hard through the period โ and gold has been running โ then a flat or even contracting ounce-denominated supply would still print a rising dollar figure. The $5.1 billion headline celebrates a metal rally as if it were chain adoption. Those are different variables with different implications, and conflating them is how a data feed becomes a narrative.
Then there is the carry problem. Holding XAUT or PAXG pays nothing. Several issuers charge custody fees on top. Compare that with tokenized Treasury products โ BUIDL, USYC and their cousins โ which are the genuine majority of the RWA category and which hand you a yield. In a world where the risk-free rate is not zero, a non-yielding gold token is a negative-carry position dressed as a safe haven.
So why does anyone hold it? For one reason, mostly: DeFi needs collateral that does not move with crypto beta. Gold's correlation to BTC is low. A lending market that accepts gold can, in theory, absorb a crypto drawdown without cascading liquidations. That is a genuine architectural contribution โ a low-beta primitive in a market that has almost none.
But it cuts both ways. In my post-mortem work across thirty protocols during the 2022 unwind, the failures that killed lending markets were never contract exploits. They were collateral assumptions. And here the collateral has an admin key. Tether has frozen addresses before. An issuer can blacklist a wallet holding tokens that a lending protocol has already counted toward solvency. That is not a code risk. That is a credit risk wearing a code risk's clothing, and it propagates straight from a Swiss vault into someone's collateralized debt position.
The competitive structure sharpens the point. KAU, at $231 million, advertises a yield-share mechanism โ the only one in the group that gestures at carry โ and holds about 4.5% of the category. The market has voted on how much it cares. The long tail, PGOLD and XAUM and a handful of unnamed others, exists mainly on listings: thin liquidity, wide spreads, and a presence in the aggregate that is symbolic rather than structural.
Meanwhile the two real contenders are diverging on opposite axes. XAUT's differentiation is Tether's liquidity and network effects; PAXG's is a New York trust charter. The market cap gap favours the former. The institutional adoption gap may favour the latter, and those two facts will not resolve quietly.
Contrarian: gold was never the flagship
Everyone frames tokenized gold as the RWA standard-bearer. The data says the opposite. Gold is 11% of the category โ a side chapel, not the nave. The $46.4 billion is overwhelmingly tokenized sovereign debt and private credit, precisely where institutions already have balance-sheet motivation and regulatory clarity. Tokenized gold is the picturesque corner of a market defined by instruments nobody photographs.
Which raises the uncomfortable question the category keeps avoiding. Traditional institutions do not need a public chain to hold gold. They have LBMA settlement, COMEX futures, and GLD at roughly forty basis points. The chain adds settlement velocity and composability. It does not add access. When you strip away the narrative, the on-chain wrapper is a distribution channel competing against channels that already work, and the same logic that makes gold tokens feel inevitable โ real asset, real backing, real audit trail โ is what makes them unnecessary for the buyers who matter most.
Artifacts of a new digital renaissance, certainly. But renaissance workshops served patrons who already had the gold.
Takeaway: what to watch through the chop
Sideways markets are for positioning, not for proclamations. Three signals will tell me whether this category is compounding or merely reflating. First, ounce-denominated supply rather than dollar market cap โ if token counts flatline while the headline grows, we are watching gold, not adoption. Second, the regulatory divergence between PAXG and XAUT; watch which one institutional DeFi actually selects as collateral once compliance teams enter the room. Third, and most decisive, the yield spread: if tokenized Treasuries keep paying and gold keeps charging fees, the collateral war is already over.
Following the thread from code to culture, the question is no longer whether real assets reach the chain. It is whether the chain earns them.
