How treasuries deploy capital onchain

Stablecoin treasury management is the practice of holding and deploying an organization's cash reserves in tokenized form: stablecoins, tokenized money market funds, and onchain lending positions. The work is sizing how much cash must stay immediately available, deciding what the remainder may fund, and evidencing both decisions to an auditor.

Key takeaways

• A US payment stablecoin issuer is barred by statute from paying holders interest or yield, so a stablecoin balance earns nothing until the treasury deploys it somewhere else.
• Tier the balance by the date the money is needed rather than by a fixed percentage split, because a 20/30/50 rule that nobody can trace back to a payables schedule will not survive its first audit question.
• The instrument's advertised redemption speed is a facility with conditions attached, not a property of the asset. Size the tier against the slow path rather than the fast one.

Just the basics

A treasury holds cash so the business can pay its bills, and invests whatever is left over so the cash is not wasting; onchain, the cash is a stablecoin and the investments are things like tokenized US Treasury funds or deposits into lending markets. The jobs are the same as they have always been. What changes is the settlement clock, running continuously rather than in business days, and the evidence trail, now a public ledger rather than a custodian's statement. The GENIUS Act, enacted 18 July 2025, prohibits a permitted payment stablecoin issuer from paying the holder "any form of interest or yield (whether in cash, tokens, or other consideration) solely in connection with the holding, use, or retention of such payment stablecoin." That single line is why a stablecoin treasury needs a deployment policy: the unit itself is not allowed to pay you.

What's in this article?

1. What is stablecoin treasury management?
2. Why would a treasury deploy onchain at all?
3. How should treasury liquidity be tiered?
4. Sizing each tier against real obligations
5. What has to be in a treasury policy?
6. Who approves a deployment, and how is it reported?
7. Failure modes, and what the treasurer sees when one starts
8. How does Railnet fit?
9. What should you do next?

What is stablecoin treasury management?

It is cash management run against tokenized instruments, with settlement on a public ledger instead of through a bank. The three obligations do not change: keep the business able to pay what it owes, preserve principal, and make the remainder productive.

What changes is the set of instruments and their behaviour. A corporate treasurer moving from a bank sweep to an onchain allocation is swapping a single counterparty with a next-day settlement convention for a set of positions with different redemption paths, different governance, and different people able to change the terms. The instruments are not exotic. A tokenized money market fund holds the same paper as its offchain equivalent. A lending market position is a floating-rate deposit whose rate is set by a published formula rather than by a rate committee.

The reason this needs a method rather than a product recommendation is that the same nominal balance behaves differently depending on the tier it belongs to. Cash earmarked for Friday's payroll and cash that nobody will touch for eighteen months are the same USDC on the same ledger; only the policy distinguishes them, and only the policy stops the second one's return target from quietly setting the first one's risk.

For the mechanics of the instruments themselves, see where onchain yield actually comes from.

Why would a treasury deploy onchain at all?

For a growing set of organizations the cash is already onchain, and moving it out to earn a return costs more than deploying it where it sits. Four situations account for most of it:

• Payments businesses and fintechs that collect and disburse in stablecoins. Moving balances to a bank to earn a sweep rate means two conversions and a settlement gap on every cycle.
• DAOs, frequently with no bank relationship available to them at all, and with reserves held by a multisig rather than an account.
• Digital asset treasury companies, whose balance sheet mandate is to hold digital assets, and for whom the question is not whether to be onchain but what the non-core reserve should do. See what a digital asset treasury company is.
• Corporates with genuine 24/7 obligations, where a Friday cutoff and a Monday value date is an operational problem rather than an accounting convention.

There is an honest counterpoint. If your receipts and payables are in fiat, your bank offers a sweep, and your finance team has no onchain operations experience, the spread available onchain is unlikely to pay for the operational surface you are about to add. Key management, reconciliation, counterparty review and a new audit conversation all have real cost, and that cost does not scale down for a small balance, so the organizations that should do this are the ones already carrying the cost for another reason.

How should treasury liquidity be tiered?

Split the balance into three tiers by the date the money is needed, and let the date determine what instruments are eligible. Do not start from a percentage.

The tiers are defined by what has to be true for the money to leave rather than by what it earns, and a position paying an attractive rate can sit in the deployable tier without contaminating anything, because nothing in the operating schedule depends on it. The same position in the operating tier is a solvency risk wearing a return.

Regulated money market funds are the closest analogue with published floors, and they are worth knowing even though a corporate treasury is not one. Rule 2a-7 requires a money market fund to hold at least twenty-five percent of total assets in daily liquid assets and at least fifty percent in weekly liquid assets, and caps dollar-weighted average portfolio maturity at 60 calendar days and weighted average life at 120 calendar days. Those numbers are calibrated for a vehicle facing daily redemption from strangers. Your treasury faces a known schedule from itself, a much easier problem. The reason to look at them is the shape: a liquidity rule that survives scrutiny states a floor, names the assets that count toward it, and constrains maturity separately.

Sizing each tier against real obligations

Build the tiers from a forward cash-flow schedule, then stress the schedule. A percentage split is the output of this exercise, never the input.

1. Build a 13-week cash forecast in the treasury's actual unit of account. If the business pays suppliers in USD and holds USDC, the forecast is in USD and carries an explicit conversion assumption, and most onchain treasury errors start here, with a forecast denominated in the asset rather than in the obligation.
2. Separate committed from discretionary. Payroll, tax, debt service and contractual minimums are committed. Marketing spend and discretionary hiring are not. Only committed items set the operating tier floor.
3. Set the operating tier to the largest committed outflow inside the horizon, plus the worst forecast error you have actually recorded. Use your own historical variance, not a round number. A treasury that has never missed a forecast by more than 8% does not need a 25% cushion, and one that routinely misses by 30% should stop arguing about yield.
4. Set the buffer to cover the gap between the operating horizon and the slowest redemption path in the deployable tier. If a deployable position takes seven days to exit under its stated notice period, the buffer must carry the treasury for at least seven days beyond the operating tier. This is the step most policies skip, and the one that turns a redemption delay into a missed payment.
5. Stress it three ways. Assume the fastest instrument is unavailable for a week. Assume one venue is fully unavailable. Assume a 5% adverse move in the unit of account against the obligation currency. Re-run the schedule under each. If any of the three produces a shortfall, the operating tier is undersized.
6. Write the resulting percentages down, with the schedule that produced them attached. The percentages are a summary of the work. Anyone who can only produce the percentages has not done the work.
7. Re-run quarterly, and after any change to the payables profile. A treasury that raises, acquires, or changes payment terms has changed its schedule, and the tiering is stale from that date.

The output is defensible in a way a fixed split never is. When the audit committee asks why 42% of the balance is deployable, the answer is a schedule, three stress runs and a notice period, not a house view.

What return is defensible on the deployable tier follows from that tier's risk budget rather than from the best rate available in the week the decision is made. If you would rather work it through against your own payables schedule, talk to the Railnet team.

What has to be in a treasury policy?

A policy that an auditor can test contains eight things: objectives in priority order, eligible instruments, concentration limits, liquidity floors, the approval chain, key and signing controls, reporting cadence, and a review trigger. Each one has to be specific enough for someone to test it.

• Objectives, ranked. Capital preservation, then liquidity, then return, stated explicitly in that order, because every hard decision later is a trade between them and the ranking is what resolves it.
• Eligible instrument list. Named instruments and named venues, not categories. "Lending markets" is not a policy; a list of specific markets with the criteria that got them on the list is.
• Concentration limits. Per instrument, per venue, per issuer, and per underlying. The last one catches the case where three positions look diversified and hold the same collateral.
• Liquidity floors. The tier sizes from the section above, expressed as minimums rather than targets.
• Approval chain. Who may propose, who may approve, who may execute, and the threshold for each escalation.
• Key and signing controls. Signer set, quorum, the process for adding and removing a signer, and what happens when a signer leaves the company, treated as part of the treasury policy rather than as an IT matter, because it is the control that actually stops a loss.
• Reporting. What is reported, to whom, how often, and in what format.
• Review trigger. Both a calendar date and a list of events that force an early review: a depeg, a governance change at a venue on the eligible list, a loss anywhere in the portfolio, a change to the payables profile.

There does not appear to be a usable crypto treasury policy template published anywhere in a form a treasurer could take to a board: every result is either a vendor's product page or a general corporate policy with the word "digital" inserted. Railnet is preparing one at /crypto-treasury-policy-template, and it will be published without a form gate.

Who approves a deployment, and how is it reported?

Five steps, and no person holds two of them: proposal, risk review, approval, execution, reconciliation. The separation is the control, and everything else is documentation.

1. Proposal. The treasury analyst writes the deployment against the policy: what tier, what instrument, what size, what concentration limit it consumes, and what the exit path is. A proposal that cannot name its exit path is incomplete.
2. Risk review. Someone outside the treasury function checks the proposal against the eligible list and the concentration limits, and confirms the instrument has not changed since it was added to the list. Governance changes at a venue are the common finding here.
3. Approval. The CFO or the treasury committee approves, at a threshold set by the policy. Above a stated size, the board or the investment committee approves instead.
4. Execution. A signer set that does not include the proposer executes. Quorum comes from the policy, not from who is available.
5. Reconciliation. A fourth party confirms the onchain position matches the approved proposal, in size and in destination, and files the transaction hash with the approval record.

Reporting is where onchain treasuries tend to underperform their offchain equivalents, the opposite of what the technology promises. The ledger is public and every position is verifiable, and yet the treasury frequently cannot produce a period-end statement that an auditor will accept without a walkthrough. The gap is that a public ledger shows transactions, and an auditor wants a position: what was held, at what value, on a stated date, reconciled to a general ledger account.

The standard the auditor is calibrated to comes from the offchain world. A registered investment company must compute current net asset value no less frequently than once daily, Monday through Friday, at times set by the board, and must price purchases and redemptions at the NAV next computed after receipt of the order. Under the custody rule for registered advisers, a qualified custodian sends an account statement at least quarterly identifying the amount of each security held and all transactions in the period, and client assets are verified by surprise examination at least once each calendar year by an independent public accountant. A corporate treasury is bound by neither rule. Its auditor was trained on both, and will ask for the onchain equivalent of each: a dated valuation, a complete transaction record for the period, and an independent verification of the holding.

Producing that on a monthly close, without a person exporting block explorer pages into a spreadsheet, is the actual operational requirement.

Failure modes, and what the treasurer sees when one starts

The losses that hit treasuries are liquidity and control failures far more often than they are yield failures, so name each one in the policy and state what the treasurer sees when it starts.

An advertised redemption path is a facility with conditions. Tokenized funds publish an instant path and a standard path, and the two have different minimums and different eligibility. Ondo's OUSG documents instant mint and instant redemption from a $5,000 minimum alongside a non-instant path with a $100,000 investment minimum and a $50,000 redemption minimum, with access restricted to accredited investors who are qualified purchasers and who have completed onboarding. None of that is unusual or hidden. It matters because a treasury that sized its buffer against the instant path has sized it against a facility that has a size limit and an eligibility condition, and both can bind at exactly the moment the treasury needs the money.

A high lending rate is often a liquidity warning rather than an opportunity. Aave's interest rate model uses two slopes with an optimal usage ratio as the inflection point, and the supply rate is derived from the variable borrow rate after the reserve factor. Above the optimal ratio the second slope raises the borrow rate steeply, and that is the mechanism working as designed: it pays new suppliers to arrive and pressures borrowers to repay. A supply rate that has jumped is therefore a signal that withdrawal capacity is tight, and it is priced precisely to discourage the withdrawal the treasurer is about to attempt. See what is an interest rate model.

Then there are dependency chains that are not visible from the position. A tokenized fund can hold other tokenized funds, and as of 24 July 2026 Ondo published OUSG's holdings as State Street's SWEEP, BlackRock's BUIDL, Franklin's BENJI, Fidelity's FYOXX, and a small USDC and bank deposit balance. That is a disclosed and reasonable structure. The treasury consequence is that a concentration limit written at the level of the instrument you bought does not measure what you are actually exposed to, and the limit has to be applied to the underlying.

Key and admin compromise comes next, and the control that fails is rarely the smart contract. It is more often a signer set that still contains someone who left, a quorum of two on a nine-figure balance, or an approval process that runs through a chat message.

Policy drift is quieter. An eligible instrument list written in January and never revisited describes venues whose parameters, oracles and governance have all moved. The review trigger exists for exactly this, and it is the clause most often ignored.

Last is the accounting surprise at close, where a position that was straightforward to enter turns out to need a valuation the auditor does not accept, or produces a taxable event the treasury did not model. Involve the auditor before the first deployment, not at the first close.

How does Railnet fit?

Railnet is the operating layer for onchain asset management. For a treasury, it is the layer that executes an approved allocation across several venues, tracks it under one state model, and produces the position record.

It is vault infrastructure, the layer that holds the strategy between the venues it allocates into and whatever channel distributes it. It is not a venue and does not compete with one. Aave, Morpho, Compound, Ondo and tokenized treasury issuers are yield sources that connect to it.

It does work at three points in that method. Execution across venues is standardized, so a deployment into four sources is one instruction rather than four integrations with four settlement conventions. Settlement is modelled in one state machine that spans both timings, so an instant lending deposit and a tokenized treasury leg sitting in a KYC gate for two days are tracked in the same standard rather than in two systems. And the position record is produced continuously rather than assembled at close, so the auditor conversation becomes a report rather than a walkthrough.

The limits are worth stating plainly. Railnet does not hold keys and is not a custodian, so the signing controls in your policy remain entirely yours. It does not choose the tiering or set the concentration limits. It enforces the mandate you write and reports against it, useful only to the degree the mandate is well written. The method itself is the part you own.

What should you do next?

• If you are drafting the policy, work through the eight components above one at a time, and watch for the template at /crypto-treasury-policy-template.
• If you are deciding what the deployable tier should hold, read where onchain yield actually comes from before reading any rate table.
• When the risk section of the policy has to survive a board, start with a risk framework for onchain allocation.
• Where the close is the blocker, read fund administration for onchain strategies.
• For implementation, the integration and reporting surfaces are documented at docs.railnet.org.

See all questions on onchain treasury management

• What is a digital asset treasury company?
• What are yield-bearing stablecoins?
• What are tokenized treasuries?
• What is a tokenized money market fund?

FAQ

What is stablecoin treasury management?

Holding and deploying an organization's cash reserves in tokenized form, and governing that deployment with a written policy. It covers the split between cash that must stay spendable and cash that may be put to work, the instruments each tier may hold, who approves a deployment, and how the resulting positions are valued and reported at period end.

Why does holding a stablecoin earn nothing?

Because the issuer is not permitted to pay you. The GENIUS Act prohibits a permitted payment stablecoin issuer from paying holders any form of interest or yield in connection with holding the token. Any return therefore has to come from deploying the balance into a separate instrument, a decision the treasury has to make and document.

How much of a treasury should be deployed?

There is no correct percentage, and any source that gives you one has skipped the work. The deployable share is whatever remains after the operating tier covers committed outflows plus your own recorded forecast error, and the buffer covers the gap between that horizon and the slowest exit path in the deployable tier. Derive it, then write down the derivation.

What is the difference between the buffer and the operating tier?

The operating tier funds obligations you have already scheduled and must be spendable with no counterparty decision involved, while the buffer funds the difference between the forecast and reality and can tolerate a queue of hours or one business day. Collapsing the two is how treasuries end up holding everything in cash and earning nothing.

Do tokenized money market funds redeem instantly?

Some publish an instant path, with conditions. Ondo's OUSG documents instant mint and redemption from a $5,000 minimum, and a separate non-instant path with $100,000 and $50,000 minimums, restricted to accredited investors who are qualified purchasers. Treat the fast path as a facility that has limits and eligibility conditions, and size the tier against the slow one.

Why is a high lending rate a warning sign?

Lending market supply rates are derived from borrow rates, and borrow rates rise steeply once utilization passes the model's optimal ratio. A rate spike means the market is paying to attract supply because withdrawal capacity is short. It is compensation for exit difficulty, arriving at the moment exit gets hard.

What does an auditor want to see?

A dated valuation of each position, a complete transaction record for the period, and an independent way to verify the holding. Auditors are calibrated on the offchain equivalents: daily NAV computation for registered funds and at least quarterly custodian statements plus an annual surprise examination for advised accounts. Expect the same three artefacts to be requested onchain.

Talk to the team

If you are sizing an onchain treasury allocation, drafting the policy that governs it, or trying to get a monthly close down from days to hours, talk to the Railnet team. There is no form to fill in first.