Public Crypto Networks as Financial Market Infrastructure

Article author: U. Bindseil, O. MalekanArticle compilation：Block unicorn

Financial instruments and processes are always designed to rely on the infrastructure they rely on. Blockchain technology, such as that adopted by public crypto networks such as Ethereum, represents a new type of payment and settlement infrastructure, giving rise to a new generation of solutions. We discuss the characteristics of the technology, such as immediacy, full asset capabilities, programmability, and the ability to flatten the financial architecture (disintermediation). We provide examples of unique products enabled by these characteristics, some of which may disrupt today's mainstream payment and capital market solutions in the future. We also discuss the utility of blockchain technology in private (permissioned) networks. Finally, we revisit the risks of public crypto networks and their mitigation measures. 1. Introduction Financial processes have always been closely linked to infrastructure. From bills of exchange to mobile banking, economic actors rely on technology and innovate over time to achieve greater efficiency. Financial infrastructures typically favor continuity—that is, uninterrupted operation around the clock—and immediacy to achieve the desired outcome—often the final transfer of funds and/or assets. In a previous paper (Bindseil and Malekan, 2024), we explored the impact of information and communication technologies (ICTs) on the temporal structure of financial instruments, payments, and settlements, and argued that technology appears to be driving finance toward a frictionless model with pervasive immediacy and continuity. Our focus is on mature technologies with higher computing power and modern communication methods such as the Internet. Our analysis also suggests that technology can only play a limited role in improving finance. At some point, the core architecture of financial interactions must evolve as well, if only to take full advantage of what the latest technologies have to offer.

This paper focuses on the distinctly different architectures of blockchain technology and public crypto networks. After describing their unique properties, we argue that together they enable temporal structures and new types of financial products that cannot exist in traditional financial architectures. Specific properties described include streaming payments, full-asset settlement, programmability, and complete disintermediation. We also explore the utility of (mostly hypothetical) private blockchain networks operated by existing intermediaries, and then provide notable examples of decentralized alternatives already operating on public networks, some with significant usage. We also consider the limitations of these solutions.

2. DeFi’s Time Structure

Compared to the current financial system, crypto networks take a completely different approach to time structure. Their unique design also stems from the fact that they were invented only recently, designed to serve a modern economy where widespread access to computers and the internet is taken for granted, and to serve a global user base. They have no “technical debt” dating back to the analog era. In other words, many existing payment and settlement systems were designed around the absence of today’s ICTs, and have to deal with spatial limitations, such as the need to deliver paper checks or certificates, or even artificial limitations, such as the slowness of manual processing. Although financial services technology has evolved, the core architecture remains the same. This is partly due to the difficulty of upgrading important infrastructure, but is also driven by incentive incompatibility. There is no technical reason why wholesale central bank payment systems could not operate on weekends, or why regular hours for stock market trading could not be extended further, but any change would require expensive adaptations for most participants and eliminate revenue streams for some.

Crypto networks are designed for an economy that demands almost global access and 24/7 immediacy. As a result, most crypto networks have little to no relationship to traditional time structures and are loosely tied to calendars and clocks. Each crypto network has its own rhythm, with the minimum interval for settlement represented by the frequency of the next block. Blocks are batches of transactions inserted into an immutable ledger by miners or validators. Depending on the protocol, block creation can be either stochastic or deterministic.

The Bitcoin blockchain targets a block time of ten minutes, but actual results are exponentially distributed and depend on several factors, including the total mining volume (Nakamoto, 2008). The protocol monitors network activity retrospectively and adjusts the difficulty of chain extensions every 2016 blocks, which is roughly equivalent to once every two weeks. Actual block times can vary from a few seconds to over an hour, but these so-called "difficulty" adjustments maintain a 10-minute average over time. Note that all other adjustments, such as the periodic reduction in the token inflation rate, are also made based on the passage of blocks rather than time. Bitcoin has probabilistic settlement finality (Lewis-Pye & Roughgarten, 2023), so the point at which a transaction can be considered "settled" is a matter of convention. But given the current high levels of mining activity - which translates into high attack costs - network participants typically consider the passage of a few blocks to be final.

Newer crypto networks have faster block times and predictive finality. Ethereum has a fixed block time of 12 seconds, with a maximum time of 12.8 minutes for irreversible settlement of any transaction (Ethereum Foundation, 2023). Other blockchains, such as Solana, offer sub-second block times and faster finality. All of these networks can be considered real-time gross settlement (RTGS) systems for payments and capital markets activities, but there are several caveats to this classification. The timing of transaction processing is never guaranteed, as whether any particular transaction is included in the next block depends on several factors, including network congestion and the fees that traders are willing to pay. The process is similar to a continuous auction without an auctioneer, as described by Huberman et al: "The Bitcoin protocol indirectly involves a priority auction, although without the presence of an auctioneer. Users with higher waiting costs pay higher transaction fees and wait less time. Users' bids have the VCG property, i.e., each user's bid is equal to the externality he imposes by delaying others' transactions." Unlike many traditional settlement systems, fees are independent of the transaction amount, but may vary depending on the complexity of the transaction. They can fluctuate and spike during periods of congestion, crowding out low-value activity. That said, crypto networks are open to the public and do not discriminate between large transfers or retail activity. They also operate continuously around the clock, with mature networks like Bitcoin and Ethereum offering close to 100% uptime.

3. Streaming Payments

The global nature of crypto networks and their continuous operation make them suitable for payments in any currency. Beyond Bitcoin, most crypto networks are designed to handle an unlimited variety of assets in a single settlement system. Networks like Ethereum and Solana allow users to issue and transfer arbitrary stores of value called tokens, extending their settlement services to assets other than their native currencies (which still require transaction fees). What any one token actually represents depends on contractual obligations (on-chain and off-chain) and social conventions. Token value is determined by the market, but can also be anchored by a promise of convertibility, including for off-chain assets.

One ​​of the most popular applications of this functionality is central bank-issued currencies, primarily the U.S. dollar. These so-called “stablecoins” are privately issued, typically backed by liquid assets denominated in the same currency (e.g., U.S. Treasuries), and are issued on various crypto networks by fintech-like companies that promise to maintain par convertibility by backing each on-chain token with off-chain reserves (which may include cash and cash equivalents). As of January 2025, there are close to $200 billion of such tokens on various blockchains, making their issuers one of the largest private holders of U.S. Treasuries. Originally invented to meet the needs of crypto speculators and as on-ramps for crypto exchanges, these instruments are increasingly used as dollar proxies for savings and payments. Stablecoin payment volume is estimated to exceed $5 trillion in 2024 (Carter 2024).

Users who trust the convertibility of stablecoins can use them for any type of payment. Unlike traditional payment systems, which are often tailored for specific types of payments (such as large transfers, retail payments, B2B, merchants, etc.), stablecoins leverage the neutrality of public crypto networks to meet a variety of needs. The main limiting factor is the temporal structure of the underlying blockchain, which leads to fragmentation. For example, due to its relatively slow block cadence and high fees, Ethereum is more likely to be used for high-value transfers that do not require instant settlement. The Tron blockchain and Solana are more suitable for small payments. As of November 2024, the median transfer amount of the USDC stablecoin on Solana is $20, while it is $1,400 on Ethereum (Ved & Cabieses 2024).

The crypto industry is working to provide faster and cheaper settlements by either improving the infrastructure of a single network or creating modular systems where low-value transfers are made on “layer 2” solutions that are settled regularly on chains like Ethereum. Whatever the approach, stablecoin payments have the potential to be instant, continuous, and virtually costless. That is, transaction costs are negligible. Unlike traditional payment providers in the private sector, crypto networks themselves make no profit. Fees are used only to prevent spam transactions and to compensate validators who protect the network. The decentralized nature of these networks means that the blockage problems and associated abuses of market power that private payment systems may eventually face do not apply.

In the future, payments on advanced crypto networks may be like streaming media, just as online media consists of streaming data. Fractional units of any currency, including fiat currencies, can flow from senders to receivers. While such payments are theoretically possible on existing electronic payment platforms, they are rarely economical due to the fee structure (which often includes a minimum fee per transaction). Even if they are economical, the intermediaries responsible for the network may arbitrarily limit or block users' abilities in order to comply with anti-money laundering/counter-terrorist financing (AML/CFT) regulations, raising reliability issues. In the crypto space, the censorship-resistant and decentralized nature of the underlying infrastructure provides additional assurances to users and entrepreneurs. This enables the latter to invent new solutions or business models that were previously impossible.

4. Full-asset infrastructure and conditional transfers

Most traditional payment and settlement systems were invented for a single currency or a specific set of securities. They also started out as regional solutions and eventually expanded to serve national or even international needs, such as international central securities depositories (ICSDs). What is often referred to as the “financial system” is actually a complex web of interconnected clearing and transfer networks, tied together by hub-and-spoke intermediaries. Interoperability is a challenge at the best of times, and can become a point of systemic risk during a crisis. Crypto networks take a different starting point. Their token capabilities transform them into full-asset infrastructures, something that traditional capital markets were not really considering. Most traditional settlement systems are designed for only a single asset (e.g., Fedwire for large-value transfers of U.S. dollars) or a class of similar assets (e.g., NSCC for U.S. stocks), or a pair of asset types (T2S for securities and euro payments). Bindseil and Malekan (2024) describe how many of these rely on large transfers and netting to achieve scale. This fragmentation is largely a byproduct of its analog roots, but has become deeply entrenched. On a purely technical basis, there is no reason to settle different assets on different networks, since all of them are database entries and all settlements are debits and credits according to the rules of the protocol. While different assets require different standards and conventions for messaging, these differences do not need to be addressed at the core settlement layer. But in reality, there are many practical, legal, regulatory, and even geopolitical reasons for different assets to be settled on different infrastructures. There is also little incentive for incumbents that enjoy a dominant market position in certain assets or currencies to want greater consolidation. Many businesses profit from fragmentation.

There are multiple benefits to settling different types of assets on the same global network. For example, market participants using different currencies do not have to wait for RTGS systems with different operating hours run by multiple central banks. An all-asset infrastructure reduces costs, reduces complexity, and reduces settlement risks such as Herstatt risk. It enables a new time structure by facilitating atomic settlement (also known as delivery-to-payment or payment-to-payment) between any assets. In traditional finance, products that need to interact with multiple settlement systems must tap into each system in turn, so that final settlement can only be completed at the speed of the slowest component. For example, a foreign securities purchase with an FX component must wait for the currency component to complete before the securities component can be settled – there is no way to parallelize the work. There are exceptions, such as the T2S platform, which is specifically set up to enable DvP in an integrated manner and settle in central bank money by including securities and central bank money in one system, but this is also a limited solution by crypto standards. The lack of commonality in settlement systems leads to frictions, such as derivatives markets that can only be settled when spot markets are open, ETFs that suffer from reduced liquidity when their constituent stock markets are closed, and securities trading that is blocked by payments. In fact, one of the barriers to moving securities markets to faster settlement is the cross-border payment bottleneck for international investors. Crypto networks do not distinguish between different assets, so a token representing an options contract can exist on the same network as the denominated currency of the transaction consideration (e.g., USD) and the spot asset of the settlement consideration (e.g., securities). Crypto networks can also sequence multiple transactions and execute them within the same block, introducing a novel form of predictability. For example, the entire lifecycle of a secured loan — including the temporary transfer of collateral — can be completed within days, hours, or even minutes. 5. Smart Contracts and Programmability The execution of such instruments is facilitated by “smart contracts,” which are conditional payments executed by the miners or validators who secure the network. Smart contracts are not legal contracts on the blockchain (Malekan 2022). They are programmed conditions that are triggered by users, other transactions within the blockchain, or external catalysts. They can be as simple or complex as any other code, and their outcomes are cryptographically guaranteed. Platforms such as Ethereum and Solana have their own virtual machines and execute code written in specific programming languages. Smart contracts that execute within a decentralized virtual machine and act on tokens are a powerful financial tool. They allow for greater flexibility in the design of financial products and more predictable outcomes at the same time. In other words, they allow extremely complex financial activities to be executed with perfect reliability. In the crypto space, a single transaction may involve 100 different parts, all of which are executed simultaneously.

An example of a new type of financial product (and time structure) that only this infrastructure can provide is flash loans. Flash loans are unsecured loans that are lent and repaid in the same transaction, giving new meaning to the concept of immediacy. They are useful for short-term activities such as arbitrage. They work by requiring the borrower to submit a list of specific activities they will do with the borrowed funds, with the final step being repayment. Paradoxically, a loan will only be approved if the smart contract providing the loan is 100% certain of repayment. This certainty may seem impossible, but it is actually how computers work. Even the most primitive calculators always return the same value when performing the same arithmetic operation. Crypto networks can also be seen as calculators that move money.

6. Disintermediation

Traditional settlement systems usually achieve efficiency by distinguishing between large and retail users. This stratification is pervasive throughout the system for four reasons: Market frictions exist, and intermediaries can overcome these frictions; Stratification allows for specialization, with some intermediaries serving end customers and others serving other intermediaries; Legacy designs are not optimized and fail to take full advantage of modern information technology capabilities; Legislation prioritizes system stability over technical optimization.  

In Bindseil and Malekan (2024), we reviewed one of the greatest efficiencies that comes with intermediation—the practice of delayed net settlement (DNS), where end-user activities are batched, netted, and settled periodically. The benefits of DNS to end-users are limited to making the intermediation of their services more efficient. But delayed net settlement comes with delays and potential risks, or costs to mitigate those risks (such as collateral for settlement risk). For example, ACH payments, which are commonly used for payroll, are typically batched on a T+1 or T+2 basis.

In a theoretically frictionless market, all participants can interact in a peer-to-peer manner and pay or settle instantly, so batching is not necessary. In practice, in such a market, any delay in settlement results in greater counterparty risk, and participants have to postpone other economic activities until final settlement is completed. Of course, batching and delayed netting have their benefits. The credit relationships created as IOUs are added and removed before final settlement make it easier for different participants to provide credit to each other, thereby reducing overall capital costs. For example, a clearing house that performs DNS can easily provide intra-settlement credit to large participants, who in turn can extend their credit to end customers such as market makers. This practice is often limited by margin requirements, but can still provide credit without external capital. The disadvantage of this approach is greater counterparty risk. In the worst case, the clearing house itself may face the risk of failure during a crisis. Margin requirements imposed to reduce this risk may trigger other disruptions, as we saw in the “Meme Stock Incident” discussed in our first paper. Real-time Gross Settlement (RTGS) systems are more explicit in terms of credit provision, and there is also a lot of leverage in DeFi. But its funding must come from external sources outside the settlement infrastructure. This increases overall capital costs but makes the system more resilient. Leverage in DeFi is easier to detect than in a clearing system that relies on DNS. In theory, peer-to-peer markets can only exist in bearer instruments - registering an asset requires some kind of intermediary in the form of a registry - and bearer assets have historically had physical form, which implies friction and risk. Overcoming this friction is one of the reasons why finance moved from bearer instruments to an IOU economy and drove the rise of ledgers (Bindseil and Malekan, 2024). Therefore, friction in almost all financial activities breeds intermediation, and most intermediaries gain both operational and capital efficiency when settlements are delayed, batched, and netted - often benefiting end users as well. This efficiency in turn enables intermediaries to better serve their customers. But it can also lead to centralization, implicit market power, technological stagnation, and systemic risk.

Crypto networks are structurally different. The assets they support are designed to be digital bearer instruments, combining the advantages of physical tokens (such as privacy and instant settlement) with the efficiency of ledger-based transfers. In theory, these networks also open up the most important infrastructure to anyone, but in practice, access is auctioned to the highest bidder, because capacity is limited, and transactions are processed in the order of the fees provided by the user. That user may be a regulated intermediary or a large user willing to pay for priority processing. Most crypto networks are designed to be unable to distinguish between different types of users in order to remain decentralized. This inability leads to less intermediation and more peer-to-peer activity, and greater immediacy for those who need it most - at least in economic terms. Peer-to-peer activity is visible through the DeFi economy, thanks to the new nature of all crypto assets as digital bearer instruments. The lack of a registry (besides the blockchain ledger that tracks everything) means that tokens - like banknotes or certificates - can be transferred from any holder to any other holder, and digitization allows frictionless transfers and near-instant settlement.  

To be fair, the financial benefits of intermediation extend beyond settlement. Intermediaries help solve the matching problem — not every lender has a ready borrower, or a buyer has a willing seller. They also pool capital, an important function for lending and liquidity provision. They also provide expertise, allocating credit in an efficient manner, and facilitate participation in financial markets by less sophisticated individuals in a secure manner. In DeFi activities where asset pooling is desirable, such as trading on decentralized exchanges or borrowing from decentralized lending solutions, smart contracts play the role of pooling intermediaries.  

Crypto networks allow snippets of code to custody assets and transfer them when predetermined conditions are triggered. This approach is not without its drawbacks, as software bugs and vulnerabilities can be catastrophic, especially when combined with the bearer nature of most tokens and the unstoppable continuity of the underlying network (see Section 10). Any decentralized finance solution where the code holds custody of a large number of assets is a honeypot for attackers. That said, pooling capital without intermediaries (in the traditional sense) is a new phenomenon in finance. It allows for scale and efficiency without the fees, delays, friction issues, and counterparty risk that typically come with pooling.  

All else being equal, fewer intermediaries means faster settlement times — assuming the underlying infrastructure can handle the volume. On blockchains like Ethereum, any user with a wallet balance can send near-real-time payments to any other user around the clock and around the world. This is in stark contrast to existing cross-border payments, which are often a coordination challenge between the sending institution, the receiving bank, and the correspondent bank. Some of these challenges can be solved without migrating to new technology, but there are challenges of governance and market inertia. These payments can only be settled at the speed of the slowest intermediary. They are also difficult to trace. On public crypto networks, buyers and sellers, or lenders and borrowers, can interact directly with code that is always running. This process eliminates the need for some of today’s brokers, exchanges, banks, and clearing houses. Fewer intermediaries means less coordination and thus faster settlement. It also means greater transparency. Disintermediation should also reduce transaction costs. Payment and settlement systems enjoy strong network externalities. Networks owned by private actors monetize their dominance through fees far above their operating costs (Rosenbaum et al., 2017). Layer 1 decentralized blockchains have only security frameworks, not business models. Still, today’s crypto networks have capacities several orders of magnitude lower than traditional payment systems, and debate is ongoing about how to scale them. One solution is to move smaller value transfers to the aforementioned secondary layers, which process faster and at lower cost, while allowing users to continuously withdraw their net worth back to the base layer. These solutions inherit some of the security, decentralization, and real-time guarantees of the underlying blockchain, but not all, giving users more choice in deciding which features they need and which they can give up. The most well-known Bitcoin Layer 2 solution is the Lightning Network. It has been operational for years but has failed to achieve significant adoption due to technical limitations and high costs. New Bitcoin Layer 2 solutions are expected to be launched in the future. Ethereum has fully embraced the layered scaling approach, and there are now dozens of such solutions, often referred to as "rollups" or L2s. As of January 2025, the most successful Ethereum L2 by dollar value of assets is Arbitrum. The most successful by transaction activity and payment fees is Base. At first glance, the layered scaling approach resembles the type of intermediation commonly seen in traditional finance. Layer 2 operators (known in the industry as sorters) perform a type of batching and give users the right to withdraw their net assets back to the main chain. Users can transfer assets from the main blockchain to the rollup by locking them in a bridge smart contract, which continuously receives activity summaries and related cryptographic proofs from the sorter. But unlike traditional clearing houses or settlement agents, well-designed layer 2s do not operate as principals or agents. Users retain the private keys that allow them to trade and withdraw assets. Crypto layer 2s are more like a code-based straight-through processing built on top of another layer - provided they optimize for decentralization. Notably, crypto users can always decide on which layer to trade on, as long as they are willing to pay the associated fees. Traditional settlement systems restrict access to the most important infrastructure to regulated intermediaries that provide batching services. 7. Permissioned Blockchains Another way to scale blockchains is to restrict the validators who maintain the network to a small group of known entities. Such “permissioned” networks can offer greater capacity than open and “permissionless” networks such as Bitcoin or Ethereum (Monrat et al., 2020). They can offer greater privacy and, because of their restrictive nature, are easier to regulate under frameworks designed for licensed intermediaries. Permissioned blockchains do not require their own cryptocurrency and have simpler consensus mechanisms. Proponents argue that they offer the best of both worlds: the benefits of tokenization, smart contracts, and composability without the technical complexity, low capacity, and resulting disintermediation of public crypto networks. Over the years, many companies have explored the benefits of permissioned blockchains. Permissioned blockchains have been used for supply chain tracking, interbank settlements, and authentication of luxury goods. The now-aborted Libra (later renamed Diem) stablecoin project, led by Meta, proposed a permissioned chain that would gradually evolve to become permissionless (Amsden et al., 2020). There are also many pilot projects conducted by central banks, such as the Swiss National Bank’s Project Helvetia and the Asian and African Central Banks’ Alliance’s Project Dunbar. The literature accompanying these projects—most of which have not yet moved beyond the proof-of-concept stage—has touted the benefits of tokenization. One example is the “unified ledger” proposed by the Bank for International Settlements (BIS), an ambitious project whose vision echoes the possibilities discussed in this article.  

Through programmability and the ability of platforms to bundle transactions (“composability”), unified ledgers allow for automated and seamless integration of sequences of financial transactions. This reduces the need for manual intervention and reconciliation that arises from the traditional separation of messaging, clearing and settlement, eliminating delays and uncertainty. The ledger also enables simultaneous and instant settlement, reducing settlement times and credit risk. Settlement in central bank money ensures the singularity of money and finality of payments.

The BIS proposal is the latest in a series of similar ideas from the private sector or public institutions, such as Utility Settlement Coin, Regulated Liability Network and Global Layer 1. All aim to improve payments and settlements by eliminating some intermediaries and empowering others. Most see blockchain as a high-value infrastructure – with public access only available through regulated intermediaries. The BIS proposal makes use of a central bank digital currency (CBDC), the integration of which the authors argue is necessary to enable the most secure types of atomic settlement. But there is no technical reason why a CBDC could not be issued on a public crypto network. A central bank – if it can accept the risks discussed in Section 9 – could issue cash-like claims on its balance sheet on Ethereum just as well as on a permissioned “unified” ledger managed by the BIS, with the added advantage of being live and battle-tested. Tokens issued on public crypto networks could still be programmed to restrict access. In addition, despite a decade of exploration, it is unclear whether any permissioned network can deliver on its promise. Most have never made it out of the pilot phase, and the few high-profile projects that have entered production (such as IBM and Maersk’s Tradelense) have shut down. The Australian Stock Exchange spent years and more than $100 million building a private blockchain to update its back-end settlement system, only to abandon it. JPMorgan claims that its internal private blockchain has processed a large number of transactions, but it can be argued that a blockchain operated by a single bank is not necessarily an efficient database setup, i.e. the same functionality can be achieved more efficiently with a centralized single ledger.  

One ​​disadvantage of any permissioned ledger is the need to codify important external considerations such as participation, ownership, and control. It is challenging to sort out the details of “who owns and controls what” between key stakeholders. Another disadvantage (and in some cases an advantage) is the lack of platform neutrality. A permissioned system can be stopped, reversed, or shut down at any time. It can also remove members’ permissions or censor their transactions. These capabilities may call into question the network’s settlement guarantees (and remind people of the importance of settlement finality). Arguably, a blockchain that can be shut down or reversed is nothing more than a complex database with a lot of unnecessary cryptographic bloat.  

Public networks do not have this problem. Ethereum cannot be stopped or reversed without the consent of a supermajority of network participants — an unlikely outcome given its large and global user base. Property rights on tokenized assets on public networks are directly tied to the fact that “no one is in charge.” Permissioned networks cannot offer similar assurances, partially undermining the original purpose of deploying such a complex and limited technology. Property rights on assets on private networks are only as strong as the contractual agreements and regulatory mandates of their operators. 8. Notable Examples The most successful crypto network to date is Bitcoin, at least as measured by the total market value of its native (and only) asset. The cryptocurrency soared to an all-time high in 2024 and surpassed a market cap of $1.7 trillion. Its blockchain settled more than $1 trillion in Bitcoin every calendar month in 2024, most of which was speculative inflows and outflows for investment purposes (i.e., not payment driven). While small compared to RTGS systems in developed economies, the Bitcoin blockchain excels in several ways. Unlike most RTGS systems, it operates continuously and can be accessed by anyone. It has also never experienced an outage or error. Bitcoin and its perceived shortcomings as an investment asset are discussed in Bindseil, Papsdorf, and Schaaf (2022) and Bindseil and Schaaf (2024). Tokens and smart contract platforms such as Ethereum and Solana also operate 24/7. Despite being less well-known than Bitcoin, these two networks routinely settle higher values ​​than Bitcoin. They also generate more revenue for their validators from transaction fees. Their smart contract functionality and ability to process transactions of an unlimited number of assets can generate high fees during periods of high demand. Today, almost all significant decentralized financial activity (beyond simple transfers) occurs on networks other than Bitcoin.

The most prominent decentralized finance (DeFi) application on Ethereum is a decentralized exchange called Uniswap. It uses a novel automated market maker mechanism to price trades according to a constant product function (Adams et al., 2020). Users who wish to act as liquidity providers (market makers) submit pairs of tokenized assets to a smart contract for pooling. Other users who wish to exchange one asset for another (takers) request quotes from the smart contract, which prices the trade based on the proportion of liquidity that will be drawn from one side of the pool. All swaps are atomic and ultimately processed by network validators. Uniswap also provides a router that finds the most economical way to process a trade, potentially routing it through multiple different pools (for different trading pairs). A simple swap of one crypto asset for another may involve four additional assets, all of which are swapped instantly and atomically. This type of trading is not possible in traditional markets, which tend to be single assets (such as stocks) and currencies (such as the euro).

As of Q4 2024, Uniswap has processed a cumulative $2.5 trillion in trading volume. There are tens of thousands of unique trading pairs, and anyone can create new trading pairs without permission. Decentralized exchanges using automated market makers have become popular in the cryptocurrency space due to the reduced computational load of processing transactions and the ability to maintain liquidity for long-tail assets (Milionis et al., 2022). Like its underlying infrastructure, Uniswap operates 24/7. Unlike off-chain cryptocurrency exchanges (such as Coinbase or the now-bankrupt FTX), users do not need to give custody of their assets to a third party to participate in trading. Therefore, it is a "non-custodial" exchange.

Other DeFi applications, such as Aave and Kamino, facilitate lending activities. Users who wish to earn interest on tokens submit them to a pool of smart contracts that hold those tokens. Other users who wish to borrow assets can withdraw them from the pool, but must first provide some form of collateral. DeFi lending is almost entirely overcollateralized, as anonymous users lack credibility and cannot seek recourse in the legal system for default. Interest rates are determined by smart contracts using bonding curves and fluctuate based on supply and demand.

Unlike most traditional banks or credit solutions, DeFi lending protocols operate 24/7 — whether loans are being originated, voluntarily closed, or involuntarily liquidated if undercollateralized. Interest accrues in real time and is calculated on a per-block basis. Some protocols have a minimum loan size, but some do not — and due to automation, the marginal cost of each new loan is zero. Aave also offers the flash loan feature mentioned earlier. Flash loans do not require collateral because repayment is guaranteed or the loan would not be issued.

Another popular activity in DeFi is the use of futures-like derivatives to make leveraged bets on the price of different crypto assets. The most popular form of such products is perpetual futures contracts, colloquially known as “perps.” Unlike traditional futures contracts, these products never expire. They use a variable funding rate charged to long and short traders, forcing the derivatives to follow the price of their corresponding spot markets. While short-term dislocations can occur, perpetual contracts should eventually return to the spot market as traders on the wrong side of the trade pay increasing fees to maintain their positions.

Perpetual futures are also a popular product on centralized crypto exchanges. In fact, there is no fundamental reason to think that such products should only be popular in crypto assets or need to exist on the blockchain. Their main advantage over traditional futures contracts is the elimination of periodic settlement and the resulting fragmentation of liquidity.

All of these DeFi solutions can be found on different smart contract and token-enabled blockchains, either as extensions of the same solution or as competitors of similar design. Ethereum is currently the go-to for high-value DeFi activity due to its history, maturity, and perceived level of decentralization. As of January 2025, its various decentralized finance solutions have over $65 billion in assets under management. Solana is a distant second with $12 billion. Solana’s faster settlement times and lower fees are offset by its history of network downtime. That said, Ethereum’s modular approach to network expansion has resulted in fragmentation across multiple interconnected blockchains, reducing the chances of atomic settlement.

Today, the vast majority of activity in DeFi serves the needs of crypto speculators. Few activities are alternatives to financial activities empowering the broader economy. This may change as infrastructure matures and new laws and regulations pave the way for greater institutional adoption. This adoption may be uneven given the needs and regulatory frameworks in different regions. But markets have a long history of migrating to infrastructure that is more inclusive, transparent, stable, and open to innovation. As a result, DeFi may attract more non-crypto activities in the future.

9. Risks and Mitigations of DeFi Networks

Public crypto networks and the decentralized applications that run on them present unique risks. The Basel Committee on Banking Supervision (2024) and the Committee on Payments and Market Infrastructures (2024; Section 4) provide a useful categorization and discussion of the risks of DeFi networks, with the latter focusing on tokenization. The risks are categorized into legal risks, network governance, technical risks, illegal activities, and settlement risks.

Legal risks arise if the legal status of a token is inconsistent with its status as an asset in its non-tokenized form, particularly in the event of default or bankruptcy. Legal issues can be particularly unclear in the context of cross-border transactions, as ideally, asset holders have a clear, enforceable legal basis for all operations in all relevant jurisdictions. In crypto networks, the direct owner of an asset can reside anywhere in the world, which is not possible in traditional finance. Another issue is the unclear intersection between the holder or part-holder nature of a crypto asset and the legal status of any asset in most jurisdictions. As the old saying goes, possession may be “9/10 of the law,” but of course, the point of property rights is to protect exceptions. Public crypto networks provide strong guarantees of cryptographic ownership (“not your keys, not your coins”), but the utility of this fundamentalist approach is not always clear. For example, what happens if the private keys of the largest owner of a tokenized stake in a large company are compromised and their shares are stolen? Should companies now recognize the thieves as the rightful owners?

Governancecan be particularly challenging for any decentralized solution that lacks enforcement power. Users of DeFi solutions on public crypto networks arguably face two layers of this risk, as both the application itself and the network it runs on are likely to have some kind of governance. Application governance is often conducted through “governance tokens,” which allow owners to vote on key decisions. This design is susceptible to low turnout, collusion between large holders, and bribery. Governance on layer 1 blockchains is often conducted informally, requiring off-chain coordination between decentralized ecosystem participants. This process can be messy, especially in emergencies. Layer 2 networks are more diverse in design, ranging from a single company controlling everything to token-based voting. Arguably, the best mitigation for governance risk is to build decentralized applications with minimal governance, like Uniswap. A mature crypto network may eventually reach a state of “ossification,” which is both stable and unlikely to change, requiring minimal governance. Beyond that, the industry will have to iterate on more effective governance models. Technically, DeFi solutions are vulnerable to code vulnerabilities, exploits, and challenges in providing off-chain data to smart contracts, colloquially known as the “oracle problem.” The fractional holder nature of crypto assets and real-time settlement of most transactions make DeFi a perpetual target for hackers. The public and censorship-resistant nature of the infrastructure makes it difficult to filter out bad actors. Finally, the decentralized nature of application governance reduces the ability of projects to respond after a breach has begun. Cumulative user losses from such issues are estimated to exceed $3 billion. While all RTGS solutions are susceptible to losing funds due to bugs or breaches, this problem is more severe in DeFi because the attack surface is larger. Nonetheless, the code of any DeFi solution can be formally verified to eliminate bugs and vulnerabilities. Projects can offer bug bounties to have others test their software, and there is a growing ecosystem of smart contract auditing firms. The largest DeFi solution on Ethereum has not experienced a major breach in years, despite having tens of billions of dollars in user assets.

The crypto networks these solutions are built on are also susceptible to uptime and consensus failure issues. The proof-of-work consensus used by Bitcoin is vulnerable to settlement reversals if a single miner controls more than 51% of the network. The proof-of-stake used by most other networks is also vulnerable to majority stake control. In addition, a PoS validator that controls more than one-third of the staked amount can delay finalization. However, these exploits have never happened on the main chain, and the parabolic rise in Bitcoin mining and the price of tokens used for staking make such attacks very expensive. Nuzzi et al. (2024) estimate the cost of a one-hour attack on Bitcoin at $20 billion, and the cost of preventing finality on Ethereum at over $40 billion.

Public crypto networks do not provide any error or exception handling — to do so would violate the aspirations of decentralization and neutrality. This design may be desirable for a native crypto asset like Bitcoin, but it is unrealistic for other digital assets. Indeed, most stablecoin issuers already have mechanisms to freeze, confiscate, or destroy tokens — the fact that tokens merely represent a claim on off-chain reserves makes this ability easy to implement. But there are no established standards for how such power should be exercised, and creating standards is more of a legal and regulatory question than a technical one. The Basel Committee on Banking Supervision defined a standard in this context according to which “all rights, obligations and benefits arising from cryptoasset arrangements are clearly legally defined and enforceable in all jurisdictions where the assets are issued and redeemed. In addition, the applicable legal framework ensures settlement finality in primary and secondary markets.” Ownership enforced by cryptographic keys also presents challenges for institutions. Many institutions are not allowed to custody their own assets for regulatory reasons. Even if allowed, institutions like pension funds cannot simply hold assets with a single key. Institutions often require full or semi-custodial services that help them manage access, succession, maker-checker operations and bankruptcy. Specialized service providers have begun to offer solutions to meet these specific needs. The public and anonymous nature of crypto usage can facilitate illicit activities such as money laundering and terrorist financing. Anonymity also poses a challenge to Know Your Customer and Know Your Counterparty (or KYC) laws in most jurisdictions. The holder-only nature of native crypto assets like Bitcoin makes them useful in ransomware attacks. Data shows that crypto assets and stablecoins are used to circumvent capital controls and evade sanctions. In oppressive political regimes, this ability of crypto can support the financing of dissidents and partially compensate for the financial exclusion that these authorities often impose on political opponents. One way to hide traces of illicit funds on-chain is to circulate them through DeFi solutions such as decentralized exchanges. The transparent nature of blockchain technology can also be used to investigate and even prevent illegal activities. Despite significant efforts to stop it, a large amount of illegal flows still pass through existing infrastructure. With regard to crypto networks, the Financial Assets Task Force (FATF) states that “without appropriate regulation, virtual assets may also serve as a safe haven for financial transactions by criminals and terrorists. The FATF has… published global, binding standards to prevent virtual assets from being used for money laundering and terrorist financing. In recent years, some countries have begun to regulate the sector, while others have banned virtual assets altogether.” A key pillar of such regulation is the “Travel Rule” (FATF Recommendation 16), which is being implemented by an increasing number of jurisdictions (FATF, 2023) and applies to so-called “Virtual Asset Service Providers” (VASPs) that engage in the transfer of virtual assets. VASPs are required to obtain and disclose detailed information related to the sender and recipient of virtual asset transfers to corresponding VASPs or financial institutions, either during the transaction or before (for transfers over $1,000). The Travel Rule is intended to bring the crypto industry into line with the AML/CFT rules of the traditional financial industry. A number of issues have arisen in the actual implementation of the Travel Rule and determine its effectiveness. In addition, payments between two non-custodial wallets are not directly covered by the Travel Rule.

Settlement finality is defined by the CPMI as the legal moment when an asset or financial instrument is transferred or an obligation is discharged that is irrevocable and unconditional and not easily revocable following the bankruptcy or insolvency of a participant. Depending on the technology and design of the crypto network, operational transfer and final settlement may not coincide, which may lead to settlement risks.

The further development of DeFi networks will need to address these challenges and authorities will need to develop effective regulation without undermining the efficient operation of crypto networks as financial market infrastructures. These potential risks also exist in principle in the traditional financial industry, but are generally mitigated through experience, investment and regulation.

10. Conclusion

Due to their unique architecture, decentralized financial applications running on public crypto networks operate at different time structures to even the most technologically advanced traditional counterparts. This architecture also allows novel products such as multi-asset atomic swaps and flash loans. Important features include the elimination of traditional intermediaries, ubiquity and flexibility of the underlying ledger, flattened hierarchies (when needed), elimination of expensive and potentially error-prone human decisions, public access, and continuous operation. These features may enable crypto networks to provide financial market infrastructure with unprecedented efficiency.

While some of these features also exist on non-blockchain infrastructures, this combination seems to be possible only on public and permissionless crypto networks. The neutrality of decentralized infrastructure provides very different assurances than the guarantees and warranties that centralized networks rely on. At the same time, various issues that have crystallized and regulated over time in traditional finance require further clarification on crypto platforms (e.g. in the areas of preventing illicit payments, settlement finality, and enforceability). Different crypto networks are suited to different types of decentralized finance, related to their speed, capacity, reliability, and user experience. Permissioned blockchains, while offering some advantages, suffer from the same organizational issues as existing settlement systems and do not reap all the efficiency benefits of public blockchains. Importantly, just because DeFi offers a certain flat design or instant time structure does not mean that market participants will choose it. Some decentralized exchanges defer settlement to maximize liquidity in periodic batch auctions. Most lending solutions use smart contracts to implement a single balance sheet that plays the role of an intermediary, albeit an automated intermediary that is largely managed by code. Indeed, some even allow owners of their governance tokens to vote on important parameters. Layer 2 solutions, in the context of maximum settlement guarantees, offer a delayed net settlement. Therefore, the biggest difference between DeFi and traditional solutions is the optionality of single-layer, instant, automated, and peer-to-peer interactions. Financial engineers are free to deviate from these extremes.

Ultimately it will be market participants who decide which time structures and product designs are optimal. The optional nature of the features described above may lead to all kinds of experimentation. Thus, the biggest beneficiaries of continued improvements in ICT may ultimately be financial engineers who will one day design new products inspired by the possible and unfettered by the past.