By Arjun Sethi, Co-CEO of Kraken
The illusion of liquidity
This happened in September 2019, when overnight repo rates rose to 10%. This happened in March 2020, when Treasury markets seized and the global benchmark “risk-free” asset went unbidden. This happened again in March 2023 when regional banks failed and the Fed had to create a new emergency facility simply to keep collateral flowing.
Each time, the diagnosis is familiar. A sudden lack of liquidity, collateral or confidence. But these are superficial symptoms. The real cause is structural: too few participants, too much focus and too much dependence on a handful of assessments.
We call these deep markets, but they are not distributed. These are highly centralized networks pretending to be decentralized. The repo market, the Treasury market, and the foreign exchange market together constitute the operating system of global finance, and that system now runs on a few machines.
We built a financial supercomputer with a single cooling fan. It works wonderfully until it doesn’t.
The architecture of concentration
Start with the repo market. On paper, this is a considerable volume, with a daily outstanding of around $12 trillion. In practice, it is dominated by four or five dealers. These same companies are involved in most bilateral trade, provide tri-party liquidity, and sit between almost all the major buyers and sellers of Treasury collateral. When a dealer hesitates, the entire chain stalls.
In the bond market, the situation is the same. A handful of primary securities specialists stand between the Treasury and the rest of the world. Market-making in Treasuries, once spread among dozens of firms, is now concentrated in fewer than ten. The buy side is not more diverse. Five asset managers control more than a quarter of global bond assets.
Trading platforms themselves, from Tradeweb to MarketAxess, have network effects that reinforce the same pattern: a small number of nodes carrying a massive volume of flows.
FX looks global, but it follows the same topology. The Bank for International Settlements estimates that the majority of daily foreign exchange transactions, about $7.5 trillion, flow through fewer than a dozen global brokers.
The interprofessional market is dense, but the customer market depends almost entirely on these same banks for its liquidity. Non-bank liquidity providers have grown, but they connect through the same channels.
In each case, liquidity appears as a property of the market. In reality, it is a property of dealer balance sheets. When these balance sheets are constrained by regulation, by risk appetite or by fear, liquidity evaporates.
We have not built markets in the form of networks. We built them as star systems, a few massive suns with all the others orbiting around their gravity.
Centralization as a feature, then a bug
This structure was no accident. It was effective when calculation, trust and capital were costly. Concentration simplifies coordination. A small number of intermediaries made it easier for the Fed to transmit policy, for the Treasury to issue debt, and for global investors to more easily access dollar liquidity.
For decades, this efficiency looked like stability. But over time, each episode of stress revealed the same fragility. The repo peak in 2019 occurred because balance sheet capacity was maxed out. The 2020 Treasury selloff occurred because the largest traders were unable to store risk. Each time, the Fed stepped in, expanding its role, building new facilities and absorbing more of the market burden.
This is not a political drift. It’s physics. The logic of centralization is getting worse. When liquidity dries up, everyone turns to the only balance sheet large enough to support the system. Every rescue reinforces the dependency.
We are now in a regime where the central bank is no longer just a lender of last resort. This is a reseller of first resort. The Treasury and the Fed together constitute two sides of the same balance sheet, one issuing guarantees, the other providing leverage against them.
The modern financial system has become a state-supported public service, not a distributed market.
Balance sheet capitalism
This is the true definition of our time: balance sheet capitalism.
In balance sheet capitalism, markets do not clear through price discovery. They compensate each other thanks to the balance sheet capacity. Liquidity is not a flow of buyers and sellers. It is the desire of some intermediaries to expand their books. The plumbing of the global system of dollars, pensions, treasuries and foreign exchange now depends on the same limited nodes.
The paradox is that each regulation intended to reduce systemic risk has worsened this concentration. Capital rules, liquidity ratios and clearing mandates all place intermediation in smaller and larger hands. The system is safer in isolation but more correlated as a whole.
When every dollar of liquidity depends on the same two balance sheets, that of the Fed and that of JPMorgan, there is no longer a market. You have a queue.
We have financialized trust into a single counterparty.
In this world, systemic risk does not come only from financial leverage. This comes from architecture. A network that appears decentralized on paper but in practice behaves like a single organism.
The more the system develops, the more its stability depends on the political and operational capacity of these central institutions. It’s not capitalism. It’s the infrastructure.
Liquidity as code
The next evolution of markets will not come from regulation. This will come from the calculation.
When you move markets on-chain, you are refactoring the system. You replace balance sheets with state machines.
Onchain markets modify three fundamental properties of liquidity:
- Transparency. Collateral, leverage and exposure are visible in real time. Risk is not a quarterly report. This is a live stream.
- Programmable confidence. Margin, clearing and settlement rules are executed by code and not negotiated by dealers. Counterparty risk becomes deterministic.
- Participation without authorization. Anyone with capital can provide or consume liquidity. Market access becomes a function of software, not relationships.
These properties transform liquidity into something structural and not conditional. It’s no longer a question of who is willing to accept your transaction. This is a property of the network itself.
Chained repo markets already exist in prototype form. Tokenized Treasury collateral, automated lending pools, and stablecoins acting as cash equivalents. The same mechanisms that govern traditional repo, collateral, margins and rollovers can be coded directly into smart contracts. Currency swaps, yield curves and derivatives can follow the same logic.
The difference is not ideology. It’s physics. It is cheaper, faster and safer to calculate trust than to regulate it.
On-chain markets are what finance looks like when liquidity stops being a privilege and becomes a protocol.
The parallel dollar system
The first real version of this world is already here.
Stablecoins are the on-chain descendants of repo collateral, dollar-denominated liabilities backed by short-term assets. Tokenized Treasury bills are the first transparent collateral instruments in financial history. And on-chain money markets, from protocol-based lending pools to tokenized repo facilities, are beginning to act as a new layer of funding for global capital.
Together, these components form a parallel dollar system, which still references the US Treasury and the Fed, but which operates according to radically different mechanisms.
In the traditional system, information is private, leverage is opaque and liquidity is reactive.
In the onchain system, information is public, leverage is observable, and liquidity is programmatic.
When stress arises, this transparency changes the entire dynamic. Markets don’t need to guess who is creditworthy. They can see it. Guarantees do not disappear into the black boxes of the balance sheet. It can instantly move to where it is needed.
The global dollar system is evolving, piece by piece, toward a public ledger. Tokenized Treasuries now exceed $2 billion in circulation and are growing faster than most traditional money market funds. On-chain stablecoin settlement volumes already rival major card networks. And as institutional adoption accelerates, these numbers will get worse.
It is no longer a marginal system. It is a mirror system; smaller, faster and more transparent than the one it discreetly replaces.
Over time, the line between onchain and offchain will blur. The strongest collateral, most efficient financing, and most liquid currencies will migrate to where transparency and composability are highest. Not because of an ideology, but because that is where the efficiency of capital is greatest.
The architecture of trust
The dollar system will not disappear. It’s an upgrade.
The financial system we built in the 20th century was centralized because the calculations were expensive. You needed hierarchies of trust, banks, brokers, and clearing houses to coordinate risk and liquidity. The 21st century system does not need these hierarchies in the same way. Verification is now cheap. Transparency is IT. Coordination can be automated.
Central banks will always exist. Treasury markets will always matter. But the architecture will be different. The Fed will not need to be the sole fan of the financial supercomputer. It will be one of many nodes in a network capable of self-balancing.
Onchain markets do not eliminate risk. They distribute it. They make it visible, auditable and composable. They turn liquidity into code, trust into infrastructure, and systemic risk into a design variable rather than a surprise.
For decades we have added complexity to hide fragility, new facilities, new intermediaries and new regulations. The next step is to remove the opacity to reveal the resilience.
What began as a speculative experiment in crypto is evolving into the next monetary infrastructure, an open, programmable foundation for global finance.
The transition will not be instantaneous. This will happen gradually, then suddenly, as all systemic upgrades do. One day, most of the world’s collateral will be settled on open ledgers, and no one will call it crypto anymore. It will just be the market.
When this happens, liquidity will cease based on who has the largest balance sheet. It will depend on who runs the best code.
And this is how finance will finally evolve from a hierarchy to a network.
