Tokenising gold

As of December 2021, global gold stock was estimated to be worth $11.9 trillion. For reference, global GDP in 2021 was $96.1 trillion.


In this week’s deep-dive, we’re going to talk about how protocols are trying to tokenise the most desired metal in the world. I choose to dig into gold because there are useful lessons for anyone who’s interested in the application of DeFi to real-world assets.

Why tokenise gold

The benefits of tokenising gold are:

  • Easier to transport: sending tokens is far easier than transporting gold.
  • Fractionalise: most of us (sadly) won’t want to buy an entire bar of gold. Users being able to denominate exactly how much gold they want to buy (e.g. $100) is an advantage.

If you wanted just the above, you could just buy a gold backed ETF. It provides exposure to gold without having to hold it. You achieve the fractionalisation by buying whatever numbers of ETF shares units you deem fit (usually 1 unit of an ETF is equivalent to an ounce of gold).

Tokenisation adds other distinct advantages on top of this:

  • Redeem your tokens for gold: gold is the ultimate store of value. It’s a precious metal and people trust it. If every other currency in the world went up, people will still trust gold. Unlike ETFs, you can always redeem your token for its value in gold.
  • Borrow using DeFi: tokenisation allows you borrow using DeFi. As I’ve written before, the underlying thesis of DeFi is to remove intermediaries and offer better lending and borrowing rates. Once your gold was represented as tokens, borrowing and lending becomes easier.
  • Better prices: protocols that tokenise gold claim to offer the best possible price. The spot price is what gold trades on the market for. Unfortunately, most private individuals end up paying a premium on this price because of intermediaries. We’ll come back to this a later.

Here’s how the stock of gold in the world is distributed. Jewellery accounts for the largest stock, followed by bars and coins and private investment.


How the tokenisation works

The objective of tokenisation is to maintain a peg between the token and the amount of gold held in a reserve.

Imagine a hypothetical blockchain with a token called $Goldchain. The blockchain’s sole purpose is to ensure that 1 token represents 1 gram of gold.


An example to bring this to life:

  1. Divya owns 10 grams of gold and wants to tokenise it.
  2. She deposits 1 kg of gold to a custodian. These custodians are responsible for storing the gold and verifying it’s quality.
  3. Once the custodian has verified the gold, 10 Goldchain tokens are issued to Divya

Once these tokens have been issued, they can be traded like any other token on the blockchain. Divya also has the right to claim 10 grams of gold using her tokens. In many ways, this is similar to a stable coin with reserves (in this case, the reserves are gold).


The flow diagram above makes tokenisation look like a simple process, but it’s more complicated. Here’s why:

Quality control

Gold comes in various forms of quality. The custodian needs to assess this quality and report it to the blockchain. Tokens need to be issued based on the quality of this gold.

Risk of theft

Custodians hold the physical asset and are responsible for its safety. The location of reserves is usually known and this makes them a target. That being said, vault operators have done this for decades. Maintaining custody of these assets is neither new or unique to the blockchain use case.


The purpose of crypto is remove reliance on a single party. In the example above, if Divya wanted to borrow against her Gold tokens, the lender will want to verify that these reserves actually exist.

This is where protocols like Chainlink can help. They provide ‘Oracle’ services. Think of these oracles as auditors who verify that a piece of information is true. They aim to validate information through multiple sources to ensure that the service is truly decentralised and they do not rely on a single party.

Do the economics add up?

Any protocol that tokenises with real-world assets deals with two main challenges. First, the cost of storing the asset. Second, proving that they actually own the asset (i.e. using “Oracles” or audits).

I dug into 3 tokens that are backed by gold: Paxos, Cache and PGMT. There are subtle differences between each token.

Paxos has a fee for tokenising gold, but charges no ongoing fees for custody. Cache charges a 0.25% custody charge, but no fees for tokenisation of gold.

PGMT is issued by the Perth Government and is backed by the gold they hold. They charge no fees for custody or transfer. PGMT is not really useful if you want tokenise existing gold assets. It’s useful if you want exposure to gold-backed tokens.

I wanted to compare these tokens to gold-backed ETFs. For simplicity, we’ll compare the Cache token, which has an ongoing charge, to ETFs. ETFs charge an annual fee, denominated as a % of the assets held, and this can be compared to Cache’s custody charge.

Cache provides is equal to or cheaper than 4 out of 9 of the top gold ETFs in the market.


As noted earlier, there are other reasons one might choose to hold a gold-backed token instead of an ETF.

Paxos and PGMT are cost efficient options if an investor only wants exposure to gold. PGMT doesn’t really offer the level of decentralisation that some way want. After all, the gold is held by a single entity which happens to be a government institution. More over, the verification of PGMT’s reserves seems to be done by themselves, not a decentralised third party like Chainlink.

If you hold physical gold and want to tokenise it, Paxos and Cache are good options. Paxos charges a fee for tokenisation and Cache charges a fee for custody.

Finally, both Paxos and Cache offer elements that an ETF does not. The gold is redeemable — if you ever wanted to exchange 10 Paxos tokens for 10 ounces of gold, you could. Do customers want this? Who knows. It doesn’t feel like something you would use everyday. But, it’s important to note that the ability to redeem may give investors some confidence.

The more interesting element here is peer-to-peer transfer. If I own an ETF, I can’t really sell it to you. I could, but it’s going to involve a lot of paper work and will likely be more expensive. I’ll need to sell my ETF (incurs a transaction fee), send you the money (assume this is free) and you convert it back into a gold ETF (incurs a transaction fee).

Even if P2P transfers with gold-backed tokens are not superior from a cost perspective, they are superior from an ease of use perspective. The status quo (ETF route) involves 3 steps, and the token route involves a single step.

To close

Real world assets on chain excite me because they unlock mainstream use cases for DeFi. Gold is the place to start - there’s lots of gold in the world, prices are well-known and people trust the asset’s underlying value.

I look forward to more assets being brought on chain. While the principles of custody and price discovery remain the same, the nuances for each asset class differ. It’s much harder to do this with real-estate, because it’s less liquid and there’s no standard, agreed price for every property in the world.

If you’re working on tokenising real-world assets, I’d love to talk to you. Shoot me a message on Twitter.