Timechain Wiki

Bitcoin banking and credit

18 min read

A Bitcoin-denominated financial layer is emerging — through Lightning Network channels, custodial services, Bitcoin-collateralized lending, proof-of-reserves-disclosed banking, and various Layer-2 protocols. The question is not whether such infrastructure will exist (it already does) but what institutional form it takes and whether it can avoid reproducing the fractional-reserve and credit-expansion dynamics the Bitcoin protocol was designed to escape. The Austrian internal debate between 100%-reserve and free-banking frameworks (see Fractional reserve banking, Free banking debate) is being relitigated by the design choices of contemporary Bitcoin financial infrastructure. The note treats the Bhatia layered-money model, the specific institutional forms that have emerged (custodial services, Lightning, BitVM, treasury vehicles), the live disputes about what Bitcoin banking should look like, and the empirical record of failures (Mt. Gox, FTX, Celsius) that has shaped community norms.

Why this note matters

Three reasons this question is load-bearing:

  1. The “Bitcoin can’t function as money without banking” critique (see Criticisms of Bitcoin) is one of the strongest mainstream objections. Engaging it requires substantive treatment of what Bitcoin-denominated financial infrastructure looks like and whether it can deliver the services modern users expect.
  2. The Austrian framework’s institutional choices apply to Bitcoin as much as to fiat. The fractional-reserve debate (see Fractional reserve banking), the free-banking debate (see Free banking debate), and the Cantillon-dynamics framework (see The Cantillon effect) all bear on what a Bitcoin-denominated financial layer should look like.
  3. The empirical record is forming in real time. Mt. Gox (2014), QuadrigaCX (2019), Celsius (2022), FTX (2022), and other custodian collapses have provided a substantial empirical base for thinking about what Bitcoin banking can and cannot safely look like. The community’s response to these failures has shaped emerging norms.

The Bhatia layered-money framework

The most useful conceptual framework for thinking about Bitcoin banking is Nik Bhatia’s “layered money” framework, developed in Layered Money: From Gold and Dollars to Bitcoin and Central Bank Digital Currencies (2021).

The historical pattern. Monetary systems historically operate at multiple layers. The base layer is the underlying monetary good (gold, base-money reserves, Bitcoin); higher layers are claims on the base layer that emerge for convenience, scalability, or service provision.

  • Gold-era example. Gold was the base layer; gold-warehouse receipts were the second layer; commercial-bank deposits redeemable in gold were the third layer; commercial-bank loans denominated in deposit liabilities were the fourth layer. Each layer extended monetary functionality at the cost of additional counterparty risk.
  • Fiat-era example. Central-bank reserves are the base layer; commercial-bank deposits are the second layer; money-market funds are the third layer; various credit instruments are higher layers. Modern monetary aggregates (M0, M1, M2, M3) roughly track the layers.

Bitcoin’s layers. Bhatia argues that Bitcoin follows the same pattern:

  • Layer 1: On-chain Bitcoin. The base monetary good; full self-custody; settlement finality after confirmation; no counterparty risk.
  • Layer 2: Lightning Network channels. Payment channels secured by Bitcoin collateral; effectively-instant payment; reduced fees; some counterparty risk during channel operation but settlement back to Layer 1.
  • Layer 3: Custodial Bitcoin services. Coinbase, Kraken, various exchange and custody services; IOUs for Bitcoin held by the custodian; convenience for the holder; counterparty risk in the custodian.
  • Layer 4: Bitcoin-denominated derivatives, lending, structured products. Futures, options, ETFs, lending arrangements; claims on claims; counterparty risk compounded.

The framework’s implication. Different financial services live appropriately at different layers. Long-term store of value belongs at Layer 1; daily payments may live productively at Layer 2; trading and speculation may live at Layers 3-4. Each user’s appropriate layer choice depends on their use case, risk tolerance, and time horizon.

Lightning Network as Layer 2

Lightning is the most-developed Bitcoin-native Layer 2. The framework matters for thinking about Bitcoin banking because Lightning channel balances are not exactly Bitcoin deposits, but they are not exactly self-custody either.

The mechanism. Two parties open a Lightning channel by depositing Bitcoin into a multi-signature on-chain transaction. They can then exchange payments by updating channel balances without on-chain transactions, with each balance update cryptographically signed. The channel closes by broadcasting the final balance state to the Bitcoin chain.

The properties. Lightning provides:

  • Near-instant settlement within the channel (no waiting for block confirmation).
  • Very low fees for in-channel transactions.
  • Privacy improvements versus on-chain transactions.
  • Scalability by moving most transactions off-chain.

The trade-offs. Lightning involves:

  • Channel-state risk — channels must be monitored to prevent counterparty cheating on closure.
  • Liquidity management — channels require inbound and outbound liquidity that doesn’t transfer automatically.
  • Routing complexity — multi-hop payments depend on intermediate nodes having adequate liquidity.
  • Watchtower dependencies — for users who can’t continuously monitor channels.

Is Lightning fractional-reserve? This is contested. Lightning channels are backed 1:1 by on-chain Bitcoin (deposited in the multi-sig); there is no leverage or money multiplication at the Lightning level. In that sense Lightning is not fractional-reserve.

However, Lightning service providers (LSPs) — companies that provide liquidity, routing, and channel-management services — can potentially operate fractional-reserve-style. If an LSP issues claims on Bitcoin liquidity it can supply without backing those claims with actual on-chain Bitcoin, it is reproducing fractional-reserve dynamics at the service-provider level.

The protocol design pushes against this, but the institutional design choices of specific LSPs determine whether the protocol’s protections are preserved or eroded.

Custodial services as Layer 3

The most-used Bitcoin financial services live at Layer 3 — custodial exchanges, wallet services, and institutional custodians. The economic question is how these can be operated without reproducing the fractional-reserve problems that fiat banking exhibits.

The fractional-reserve risk. Custodians hold customer Bitcoin in pooled storage. Customer accounts are accounting entries against this pool. If the custodian lends out customer Bitcoin to traders or to other counterparties, it has effectively issued more claims on Bitcoin than it holds in reserve — fractional reserves at scale.

The empirical record. Multiple major custodian failures have exhibited this pattern:

  • Mt. Gox (2014). Held approximately 850,000 BTC of customer deposits; lost approximately 740,000 through some combination of mismanagement and theft. The collapse exposed that most exchanges did not maintain segregated customer assets.
  • QuadrigaCX (2019). Canadian exchange that lost approximately $190M in customer funds. Investigation revealed the founder had been operating effectively as fractional reserves, with customer assets being used for personal trading and investment.
  • Celsius (2022). Crypto lender that promised yields on deposited Bitcoin. Collapsed in mid-2022 when the leverage embedded in the yield structure unwound. Customer funds were inadequately segregated.
  • FTX (2022). Major exchange that collapsed in November 2022 amid revelations that customer funds had been used to backstop trading operations at affiliated Alameda Research. Effectively a fractional-reserve operation that became insolvent when customers attempted withdrawals.

Proof-of-reserves as response. Since 2022, leading exchanges have implemented proof-of-reserves protocols that allow cryptographic verification that the exchange holds sufficient Bitcoin to back customer deposits. Specific implementations vary:

  • Merkle-tree proof-of-reserves. Customers can verify their individual account is included in the exchange’s claimed liability list; the total liability is cryptographically committed; the exchange separately proves it holds Bitcoin equal to the claimed liabilities.
  • Proof-of-liabilities (limitations). The harder problem is proving all liabilities are disclosed — a custodian could fail to include customer accounts in the liability list, understating its obligations. This is partially addressed by inclusion proofs but remains incomplete.
  • Proof-of-solvency. A complete framework requires both proof-of-reserves and proof-of-liabilities — currently no standard satisfies both perfectly.

The proof-of-reserves movement is the closest thing the Bitcoin community has to an institutional response to the fractional-reserve problem. It is incomplete but substantially better than the pre-2022 state.

Bitcoin-collateralized credit

A separate Layer 3+ infrastructure is emerging around Bitcoin-collateralized lending — using Bitcoin as collateral for fiat or stablecoin loans without selling the underlying Bitcoin.

The use case. Bitcoin holders who don’t want to sell (avoiding capital-gains tax, maintaining long-term exposure) but want access to liquidity can post Bitcoin as collateral for loans. The loan-to-value ratio (LTV) is typically 30-50%, providing substantial collateral cushion against price volatility.

The institutional forms.

  • Direct custodial lending (BlockFi, Celsius pre-collapse, various others) — the customer transfers Bitcoin to the lender, who custodies it and lends fiat against it. Highest counterparty risk; the historical collapses have mostly come from this model.
  • Multi-signature collateral arrangements (Unchained Capital, Casa) — the Bitcoin is held in multi-sig vaults with the borrower retaining one or more keys; the lender cannot unilaterally move the collateral. Substantially lower counterparty risk than direct custodial lending.
  • DeFi-style automated lending — various Layer-2 protocols and sidechain implementations (Stacks, Sovryn, others) that use automated smart contracts to manage Bitcoin-collateralized loans without trusted intermediaries.

The fractional-reserve question. Bitcoin-collateralized lending is not necessarily fractional-reserve — if the lender holds the Bitcoin in reserves and lends fiat against it, the Bitcoin side is fully backed. The lender’s fiat-side exposure is what creates risk.

However, if the lender rehypothecates the Bitcoin (uses it as collateral for the lender’s own borrowing), it has effectively issued claims on Bitcoin that can exceed the underlying collateral. Rehypothecation is the specific mechanism by which the Celsius collapse occurred.

The conservative practice. Lenders that explicitly commit to not rehypothecating customer collateral (Unchained Capital, various others) operate effectively as 100%-reserve institutions. This is the framework most aligned with the Rothbardian Austrian position.

BitVM, vaults, and protocol-level extensions

Recent protocol-level developments enable additional banking-style functionality without reproducing the historical fractional-reserve problems:

BitVM. A 2023 proposal (Robin Linus, et al.) for general-purpose computation enforceable on Bitcoin without protocol changes. BitVM enables more complex contractual arrangements (escrow, conditional payments, automated lending) at the Bitcoin Layer 1 level. The framework is early but potentially enables institutional arrangements that don’t require trusting custodial banks.

Vaults. Proposed protocol-level features (BIP-119, BIP-345, various others) that enable time-delayed Bitcoin movement, key-rotation, and recovery mechanisms. Vaults provide bank-like security properties (delayed withdrawal, multi-party authorization) without delegating custody to a bank.

Sidechains and federated systems. Liquid (federated, by Blockstream), Rootstock (federated Bitcoin sidechain), and various others provide additional functionality through trusted-federation models. These represent a Layer-2.5 — more centralized than Lightning, more decentralized than custodial services.

The institutional implication. Protocol-level developments may eventually obviate much of the institutional banking need. If individual users can have bank-grade security (multi-sig vaults), bank-grade convenience (instant payments through Lightning), and bank-grade credit access (BitVM-enabled or sidechain-enabled lending) without trusting a centralized institution, the role of Bitcoin “banks” may be substantially smaller than in fiat banking.

The free-banking question applied

The Austrian internal debate between Rothbardian 100%-reserves and Selgin-White free banking (see Free banking debate) directly applies to Bitcoin’s emerging financial layer.

The Rothbardian-aligned position. Bitcoin-denominated banking should be 100%-reserve by design and culture. Proof-of-reserves should be a minimum standard. Rehypothecation should be opposed. The “not your keys, not your coins” framing should remain dominant. Custodial services should be regarded with appropriate suspicion.

This is the position represented by:

  • The strong-maximalist Bitcoin community
  • Caitlin Long’s Wyoming-banking framework (regulatory framework requiring 100% reserves)
  • Andreas Antonopoulos’s emphasis on self-custody
  • Unchained Capital and Casa’s multi-sig-vault approach

The free-banking-aligned position. Bitcoin-denominated banking will inevitably develop fractional-reserve characteristics, and this is acceptable under appropriate market discipline and transparency requirements. Competitive Bitcoin-denominated banking with proof-of-reserves and well-designed institutional structures can provide the services modern users want while maintaining adequate safety.

This is the position represented by:

  • Allen Farrington’s institutional-civilizational framework
  • Various Bitcoin-banking startups attempting to combine Bitcoin custody with fiat-banking services
  • Some readings of Bitcoin is Venice (Farrington and Meyers)

The unresolved question. The empirical record of 2014-2026 substantially favors the Rothbardian-aligned position. Every major Bitcoin custodial failure has been fractional-reserve-style; transparent fully-reserved alternatives have not failed; the cultural pressure toward self-custody has strengthened over time. But the empirical record may be reading specific implementation failures rather than the underlying institutional question.

The next decade should provide more evidence. If proof-of-reserves-verified fractional-reserve institutions operate stably, the free-banking position will gain support. If failures continue at the rate of 2014-2022, the Rothbardian position will be reinforced.

The civilizational implications

A Bitcoin-denominated financial layer would have implications beyond the technical-institutional questions. Allen Farrington’s Bitcoin is Venice framework develops this dimension:

Capital allocation under sound money. Under fiat regimes, capital allocation is distorted by Cantillon dynamics — credit flows to those positioned to receive it first, not to highest-productivity uses. Under sound money, capital allocation would be driven by genuine time preferences and productivity differentials. The institutional structures supporting this allocation are what a Bitcoin-denominated financial layer would need to provide.

Time horizons. Sound money lowers time preference (see Time preference and money, Low time preference as civilizational virtue). Longer time horizons enable longer-term investment, infrastructure projects, intergenerational wealth-building, and broader civilizational development. The Renaissance-Venice framework (Farrington) is one historical analogue for what such a regime could produce.

Decentralized institutional capacity. A Bitcoin-denominated financial layer would, if it follows the protocol’s institutional logic, be structurally more decentralized than fiat banking. Decentralized institutions have different failure modes (less systemic, more idiosyncratic) and different scale dynamics (smaller, more numerous institutions). This implies institutional ecology different from the fiat-banking concentration we observe.

The honest reading. The civilizational case for Bitcoin-denominated finance is speculative — there is no historical equivalent of a genuinely-sound-money modern economy to point to. The case rests on theoretical extrapolation from Austrian framework rather than direct empirical evidence. The next several decades should provide that evidence one way or the other.

Counter-arguments and tensions

The “this won’t scale” objection

The argument: Even granting that Bitcoin-denominated banking can work for committed Bitcoiners, scaling to mainstream adoption requires services (instant payments at consumer-internet scale, credit cards, loans at the speed and scale of fiat banking) that Bitcoin’s protocol simply cannot support. The “Bitcoin banking” framework is a niche solution, not a replacement for the modern financial system.

Response: Partially right. Bitcoin Layer 1 cannot scale to fiat-banking transaction volumes. The question is whether Layer 2 (Lightning), Layer 3 (custodial services with proof-of-reserves), and protocol-level developments (BitVM, vaults) can collectively provide the scalability. The empirical evidence suggests this is at least possible — Lightning’s transaction volumes have grown substantially — but the question is unresolved. The Bitcoin-banking proposition may end up serving a smaller fraction of total financial activity than fiat banking does, with implications for what mass adoption looks like. The scaling question itself is engaged substantively at Network capacity and fee-market critiques and Lightning Network operational critiques.

The custodial-centralization objection

The argument: In practice, Bitcoin banking will likely concentrate in a small number of large custodians (Coinbase, Fidelity, BlackRock, etc.), reproducing the centralization of fiat banking even within a decentralized-protocol monetary good. The promise of decentralized Bitcoin banking is theoretical; the reality is concentrated custodial services.

Response: Real concern. The custodial-services market has shown substantial concentration tendencies — network effects, regulatory compliance costs, and security infrastructure all favor large incumbents. The honest reading is that Bitcoin banking will probably be more concentrated than self-custody advocates hope but less concentrated than fiat banking has become. The structural pressures cut both ways, and the equilibrium will depend on regulatory choices, cultural norms, and protocol-level developments. The substantive treatment of the institutional-concentration trajectory — and why the self-custody exit bounds it — lives at Custody concentration risks.

The “Bitcoin banking is just dollar banking with extra steps” objection

The argument: Most current Bitcoin “banking” services are denominated in dollars, hold dollar-denominated reserves and liabilities, and offer Bitcoin only as an asset class within a broader financial structure. Calling this “Bitcoin banking” overstates the case; it is actually fiat banking with Bitcoin exposure.

Response: Largely correct for the current state. Genuinely Bitcoin-denominated banking — where the unit of account, the liabilities, and the credit operations are all in BTC rather than USD — is rare in 2026. The Bitcoin-as-unit-of-account world remains future. Most current “Bitcoin banking” is fiat banking that happens to custody Bitcoin. The framework matters for thinking about what genuinely-Bitcoin-denominated banking would look like, but the current empirical record is largely about Bitcoin-as-asset, not Bitcoin-as-currency.

The regulatory and political-economy challenge

The argument: Whatever Bitcoin banking looks like technically, the regulatory regime determines what actually exists. Banking regulation, anti-money-laundering rules, securities regulation, and tax treatment all shape the institutional landscape. The “Bitcoin banking” framework underestimates how much the institutional outcome is determined by regulation rather than protocol design.

Response: Real and substantial. The political economy of Bitcoin regulation is at least as important as the protocol-level design. The Austrian framework’s relative inattention to political-economy of its own institutional alternatives is a real gap (see Critiques of Keynesian economics). The Bitcoin community’s engagement with regulatory questions has been uneven — strong on some questions (resistance to overreach), weak on others (proactive design of acceptable regulatory frameworks). This dimension deserves more substantive treatment than the framework has produced.

Taken together, these tensions describe the near-term reality honestly: most 2026 “Bitcoin banking” is dollar-denominated, custodially concentrated, and shaped as much by regulation as by protocol. What none of them touches is the direction. Each is an argument about how far the transition has traveled, not about whether the destination — credit and banking that settle in a money no issuer can debase — is reachable; and the property that makes it reachable, an exit to self-custody that stays open at every layer, is exactly what fiat banking cannot offer its depositors. Bitcoin-denominated banking is an underbuilt frontier, not a foreclosed one.

Open questions for further development

  • Will Lightning Network successfully scale to mainstream payment volumes, or will it remain a niche infrastructure for specific use cases?
  • Will Bitcoin-collateralized lending develop into a substantial credit market, and if so, will it preserve the 100%-reserve framework or develop fractional-reserve characteristics?
  • How will the regulatory regime for Bitcoin banking evolve, particularly in the US, EU, and major Asian markets?
  • Will BitVM and protocol-level developments substantially reduce the need for institutional Bitcoin banking, or will custodial services remain the dominant mass-adoption pathway?
  • What is the appropriate institutional structure for Bitcoin treasury vehicles (Strategy, various corporate treasuries, sovereign positions), which are large-scale Bitcoin custody arrangements with different incentive structures than retail-facing banks?
  • Will the Bitcoin community successfully resist the institutional pressure toward fractional-reserve banking, or will those pressures eventually prevail as adoption scales?
  • How does the Bitcoin-banking framework interact with stablecoins (USDT, USDC, others), which are de facto fractional-reserve dollar instruments built on blockchain infrastructure?

Canonical sources for this note

Layered money framework

  • Layered Money, Nik Bhatia (2021) — canonical framework
  • Various Nik Bhatia writings and podcasts extending the framework

Lightning Network

  • Mastering Lightning Network, Antonopoulos and Osuntokun (2021) — technical reference
  • The Lightning Network, Joseph Poon and Thaddeus Dryja (2016) — original whitepaper
  • Various Lightning Labs and Acinq technical documentation

Bitcoin banking and custody

  • Bitcoin is Venice, Allen Farrington and Sacha Meyers (2022) — civilizational framework
  • Various Caitlin Long writings on Bitcoin-denominated banking (Wyoming SPDIs)
  • Coinbase, Kraken, Fidelity Digital Assets institutional research
  • Unchained Capital and Casa custody framework documentation

The custodial-failure record

  • Mt. Gox litigation documents and Wikipedia summary
  • QuadrigaCX bankruptcy proceedings and Ernst & Young reports
  • Celsius bankruptcy proceedings
  • FTX bankruptcy proceedings and Sullivan & Cromwell reports
  • Various Bitcoin Magazine and CoinDesk reporting on each episode

Proof-of-reserves

  • Greg Maxwell’s 2013 initial proof-of-reserves proposal
  • BitMEX Research papers on proof-of-reserves implementations
  • Nic Carter writings on proof-of-reserves practice
  • Various exchange documentation (Kraken, Coinbase, Bitstamp implementations)

Austrian institutional framework

  • The Mystery of Banking, Murray Rothbard (1983)
  • The Theory of Free Banking, George Selgin (1988)
  • The Ethics of Money Production, Jörg Guido Hülsmann (2008)
  • Denationalisation of Money, Friedrich Hayek (1976)

Bitcoin-side critical perspectives

  • Andreas Antonopoulos, various talks on self-custody and the limits of custodial services
  • Jameson Lopp, security and custody analyses
  • Bitcoin Optech newsletters covering custody and Layer-2 developments

Mainstream financial-economic engagement

  • Various Federal Reserve research papers on cryptocurrency and stablecoins
  • BIS reports on crypto financial-stability implications
  • IMF working papers on Bitcoin and crypto-asset banking

Graph View

Backlinks