原文：《Deconstructing JP Morgan's "Institutional DeFi" Forex Transaction: A Case Study》by CollinSellers
A few months ago, JPMC conducted what they called “DeFi Institutional” transactions. Let's break down what happened, why it matters, and what it might mean for the future of capital markets.
It feels like there's a new headline vying for my attention and analysis every week. Whether it’s BlackRock’s Larry Fink touting asset tokenization as a “next-generation market” or State Street’s optimistic statement of a “significant opportunity” for asset tokenization , or BYN Mellon CEO’s comments on the need for Bank of New York Mellon to participate in digital assets, large financial institutions are taking notice, gathering market intelligence, building products, and laying the groundwork for blockchain-based capital markets infrastructure. Base.
JPMorgan is one of those players, specifically their blockchain division Onyx, which caused some incredible buzz a few months ago when they announced they had successfully completed a foreign exchange transaction.
I remember seeing these headlines and being intrigued by what they were doing. Then last week, more headlines appeared with a report written by Oliver Wyman and JP Morgan, and I parsed through the public information I could find to make sure I had as complete an understanding as possible of what they were doing.
Full disclosure, I only have public information to work with (I would love to interview one of their team members more in depth about this), so there may be gaps between my understanding and the reality of what’s happening, but to the best of my journalistic ability, based on what I could find about the Onyx team, interviews, and comments, here is my analysis of their “DeFi Institutional” FX trade.
Public blockchains, permissioned pools, and protocols
After the deal closed, Onyx’s division head Tyrone Lobban made a series of lively comments in interviews and podcasts about how JPMC should approach the world of blockchain infrastructure. I was impressed by how clearly he articulated the challenges facing large financial institutions and how JPMC might address those specific issues.
From an adoption perspective, private blockchains are difficult to get off the ground, they require ongoing maintenance by participants, and are highly subject to centralization, effectively negating the core concepts of decentralization and transparency. Almost every blockchain attempt I have seen that promised "enterprise" or "institutional-grade" functionality has failed. Beyond that, private blockchains are really just private databases that are subject to the same security risks as all technology-enabled companies when using typical cloud database infrastructure.
A note on private permissioned blockchains: In the eyes of the law, private permissioned blockchains are essentially synonymous with private databases, which have been permitted for securities trading for decades, particularly after the DTCC (U.S. Central Securities Depository) launched the Direct Registration System in 2006, which made “digital databases” widely available to capital market participants, particularly brokers, transfer agents and stock exchanges.
I disagree with the private permissioned blockchain approach because it strikes me as two-faced and leaves market participants vulnerable to fraud. On one hand, you are trying to immediately establish trust and credibility by simply using the word “blockchain,” but on the other hand, it is a completely controlled environment where the blockchain architect has full control over everything that is written and recorded. My concern about using a private permissioned blockchain is that it is technically easy for a fraudulent player to roll back the entire chain of transactions to remove a transaction, legally or illegally.
The ability to roll back a transaction to correct a mistake or erase the history of a transaction is a world away from an open blockchain. For this reason, I believe private permissioned blockchains will not be as impactful as public permissionless blockchains, which are supposed to keep the ledger historical and immutable at all times, barring any catastrophic crash or operational hiatus. I see no reason why large financial institutions could not use public blockchains with private permissioned protocols to get the benefits of a controlled environment without the risk of fraud in terms of historical and immutable data.
Onyx conducted this transaction on Polygon, an EVM-compatible sidechain, to keep transaction fees and latency low, but beyond that, the environment they built is a permissioned one, with the protocol itself only allowing access to transactions to whitelisted traders. In other words, instead of using a private blockchain, Onyx chose to use a public blockchain, but a private permissioned protocol with on-chain identity and other controls to make the transaction possible and compliant. See Aave Arc, a permissionless protocol, to learn more about the combination of a public blockchain and a private permissioned protocol.
At its core, this is a simple forex transaction where two parties hold different denominations of currency and collateralize one currency to another through the Aave Arc protocol. The important differences to note between this transaction and a standard forex transaction are as follows:
First, Aave is a lending protocol, not a decentralized exchange. This means that SBI and JPMC each pledged their assets (tokenized currency deposits) and received each other’s assets in exchange. In the DeFi world, you typically pledge your asset — ETH, for example, and in return you receive a liquidity pool (LP) token that provides a yield to the holder as long as they hold ownership of the LP token and put their collateral asset (ETH) in the pool for lending. This is the same thing, except instead of ETH and LP tokens, it’s tokenized Japanese Yen and tokenized Singapore Dollar.
Next, Aave is not an FX backing protocol involving fiat currencies. They are a decentralized lending platform and open source liquidity protocol for digital assets and cryptocurrencies. As far as I know, Aave is not a money transmitter, nor do they have any method for transferring fiat from one institution to another, which makes this transaction a huge leap forward in creating an entirely new market using their technology, compliant with regulations.
Due to fiat currency restrictions that Aave cannot circumvent, JPMC’s transactions are tokenized deposits, not stablecoins or fiat currencies themselves. This distinction, while subtle, is absolutely critical, as other capital market participants, such as Figure (a financial services company using blockchain technology), have also taken a similar approach to executing transactions on-chain, that is, tokenized instruments as liabilities of banks, rather than stablecoins like USDC, which are held in custodian banks and backed by paper fiat currencies.
Rather than trading between GBP and USD, it’s more about tokenizing SGD and JPY.
Tokenized Currency vs. Stablecoin
At the Duke Digital Assets event last month, USDF Alliance CEO Rob Morgan made several statements that felt aligned with Onyx’s position, particularly regarding the governance and full backing of stablecoins with the flexibility of tokenized deposits that can serve as bank liabilities and support credit creation.
This gets to the heart of the fractional reserve system, the flaws of which I will leave entirely aside in this analysis, but I want to emphasize that for a traditional stablecoin like USDC, every USDC coin in circulation must be backed 1:1 by actual dollars in an audited bank vault, whereas with tokenized deposits, banks can create credit on top of depositors’ assets as liabilities of the bank. This difference between tokenized deposits and stablecoins is also the primary method by which many of these proprietary payment systems, like JPM Coin, Figure, or Signature Bank, are able to run transactions 24/7 rather than being bound by the traditional time constraints of the modern banking system.
Just last week (as of this writing), Oliver Wyman and Onyx released a research report examining the value of tokenized deposits as a method of facilitating instant or instantaneous trading, intraday settlement, and collateralization.
From this diagram, we understand that JPMC's approach is to build blockchain in tandem with existing banking infrastructure, rather than seeking to completely replace it. I've said it before, and I'll repeat it here: I believe the future of capital markets will be a hybrid combination of on-chain/off-chain systems that allow for faster settlements but maintain existing fiat currencies and compliance channels.
Transaction process
Since Aave is a lending protocol, these tokenized deposits are explicitly intended to be borrowed by whitelisted and approved counterparties, with SBI providing collateral in tokenized Japanese Yen and Onyx providing collateral in tokenized Singapore Dollars.
The basic process can be broken down into the following points:
Keep in mind that tokenized deposits are a liability for the bank. In a podcast detailing the deal, Lobban called tokenized deposits “effectively debt instruments,” but they can serve the purpose of maintaining liquidity in the modern banking system, unlike reserve-backed stablecoins, which typically must be backed by the stablecoin issuer’s own reserve mandate. This required reserve backing can both make credit creation difficult and create liquidity issues for the system as a whole, something Lobban has also mentioned in several interviews since the deal closed.
With that in mind, here’s another look at the visuals:
This is the general idea of the transaction - using the Aave (not a foreign exchange provider) protocol to support asset lending between two counterparties. It goes through the Aave protocol, pledged and borrowed by both parties, and is not settled bilaterally. Of course, it can also be achieved through decentralized exchanges such as Uniswap.
On-chain ID (identity) management
In order to properly facilitate this transaction, a system must be built to link the identities of these wallets on-chain. Keep in mind that most public blockchains that support asset tokenization are pseudonymous, not anonymous, which means that with the proper infrastructure and tools, you can confirm and maintain the identity of the wallet owner. In order to perform this transaction, Onyx splits the credentials into two components, both of which are required to proceed:
1. The ID of the entity (such as Vertalo)
2. Individual trader ID (e.g. Collin Sellers)
Onyx has gained support for identity certificates, called “verifiable credentials” to manage the identities of entities, as well as the identities of individual traders. Lobban has mentioned on multiple occasions that this part of what they have built is something they are very proud of.
The ID product is also interesting because its permissions are designed to be managed at the user level, rather than by a government or central entity. Specifically, it was noted that Onyx:
1. Do not write personally identifiable information (PII) on-chain, and
2. They only allow “programmability” of IDs to be set by the end user, preventing the weaponization of the technology against the end user.
This attention to detail does concern me because on-chain identity will be an important factor in blockchain-based capital markets, but it also poses a danger to privacy, property rights, and freedom of exchange if it is used against you, especially as an extralegal weapon.
I am glad to hear that Onyx is highly aware of this and is working to incorporate controls into its systems to prevent identity technology from being used against traders, corporate entities, or others. For centuries, the United States has led the global capital markets precisely because individuals, businesses, and even governments have stable and reliable property rights both domestically and internationally, and I believe we must keep this top of mind as we build new markets using emerging technologies such as big data, blockchain, and artificial intelligence.
Public protocol, permission access
One thing I heard repeatedly from the Onyx team was their insistence on not asking Aave to modify their existing smart contracts and pooling protocols, but rather asking Onyx to enter the Aave universe as is, and only apply changes outside of the protocol when JPMC deems it necessary. This hybrid approach is one I expect to see more of from large financial institutions in the future.
Taking into account regulatory constraints, here is the process divided into several steps:
To keep this diagram simple and understandable, I have omitted JPMC and SBI sending their tokenized deposits into the Aave Arc protocol, and instead show the permissioning process of directly entering the Aave protocol, along with the bank liabilities held by each counterparty.
But let’s remember again that everything before Aave was managed entirely by Onyx to support compliance, banking standards, and their own internal requirements, which means it actually looks like this:
This was performed on a public blockchain, through a private permissioned pool, using a private permissioned protocol, supporting foreign exchange trading in currency pairs using tokenized deposits as bank liabilities, which is quite impressive.
Transactions do not generate Gas (transaction fees)
The last point to note here is around the transactions themselves, which are paid for using Gas. Due to Basel III restrictions, as well as internal standards around regulation and compliance, JPMorgan as a bank cannot hold or custody cryptocurrencies, which presents a challenge when paying transaction fees to the Polygon network; in this case, the transaction cost about $18 to execute.
Strictly speaking, it’s not that banks under the supervision of the Bank for International Settlements “can’t” hold cryptocurrencies, but rather that the risk weight for holding cryptocurrencies in bank custody is 1,250% as of December 22, which is very heavy and difficult to handle from a balance sheet management perspective. JP Morgan Chase CEO Jamie Dimon has never been shy about expressing his disdain for the entire crypto industry, so it’s not surprising that JP Morgan as an institution chooses not to hold any cryptocurrencies (yet). (Block unicorn note: Because every transaction on the blockchain network requires a transaction fee, ETH is used as a transaction fee on the Ethereum network, and MATIC tokens are used as transaction fees on the Polygon network.)
To meet the needs of their situation, Onyx hired a company called Biconomy, which has the concept of "Gas-free transactions", allowing JPMC to create wallets but not host them, with the express purpose of not having to pay for Gas, and then write permissions on the chain to dictate who can use that wallet when paying for network transactions. This is also called a "meta transaction", and it is more widely used in the DeFi field. From Biconomy, we can see the definition of a meta transaction, making the transaction "Gas-free".
How are Gas-free transactions achieved?
Sending a meta-transaction is similar to sending a standard transaction (from sender to recipient, and then to signature confirmation transaction), but instead of sending it directly to the blockchain, the meta-transaction is sent to a third party, who handles the Gas.
The intermediary is not Onyx, but an independent third party that takes care of the cryptocurrency and handles the gas fees before uploading the transaction to the blockchain. I must say that using meta transactions to achieve "no gas" is both simple and brilliant given how strict the JPMC's compliance standards are.
I would like to see other banking institutions develop symbiotic relationships with gas providers or digital asset custodians to support future smart contract licensing transactions like this one.
Summarize
Onyx’s “DiFi Institutional” transactions are unique in that they allow large financial institutions to execute DeFi licensed transactions on-chain, a process that includes: public blockchains, licensing protocols, licensed liquidity pools, Aave Arc, and managing full compliance components, on-chain identities for entity IDs and trader IDs, gas-free (meta) transactions, tokenized bank deposits (not stablecoins), and foreign exchange transactions utilizing lending protocols.
As we see more and more organizations building products around blockchain, I hope to see other players take similar approaches and hope that they can learn from the process, apply the lessons of successful examples, understand the various nuances, and use these experiences to build decentralized applications on the blockchain.
