North Dakota’s bank coin runs on Solana. Follow the dollar back
Bank of North Dakota’s Roughrider Coin is live on Solana as part of a Fiserv digital asset platform announced October 1. More than 90 banks and credit unions can access the system, according to Fiserv. The headline sounds like a state bank issuing a coin on a public blockchain. The actual chain of responsibility is longer: a nationally chartered bank issues it, another bank governs the program, a payment technology company handles the bank interface and a wallet provider supplies infrastructure.
- Fiserv announced the live Roughrider Coin use case on October 1, with access for more than 90 institutions.
- VersaBank USA issues, mints, burns and manages reserves; Bank of North Dakota provides governance oversight.
- The coin is permissioned on public Solana and restricted to financial institutions, not retail investors.
- Each participating institution has a VersaBank custody account; daily flows connect to a concentration account at BND.
- The state’s described autoburn design makes the fiat settlement and reconciliation record as important as the onchain transfer.
In the Fiserv announcement, the platform is live with financial institution clients and Roughrider Coin is its first use case. Fiserv calls it a dollar-backed stablecoin. The Bank of North Dakota’s own operating description calls it a token deposit for financial institutions. Those terms can carry different legal and economic expectations; the operational facts should come before the label.
The hard question is what a recipient bank gets after a token crosses Solana. A token transfer records an authorized change in the digital asset’s state. It does not itself identify the bank account holding the backing dollars, clear an ACH file or answer which entity must make the holder whole if the issuer fails. The state bank’s detailed description supplies a route through designated accounts, automated mint and burn, and daily netting. That route is the story.
Four names appear on one transaction
Bank of North Dakota, or BND, sponsors the use case and says it provides governance in an oversight role. VersaBank USA National Association, an OCC-chartered U.S. bank, issues the asset and provides minting, burning, custody and reserve management. Fiserv runs the platform and the Commercial Center interface used by participating institutions. Fireblocks provides secured wallet and tokenization infrastructure. Solana processes the onchain transaction. A reader can assign each role without making the state bank the issuer or the chain the holder of the dollars.
Fiserv’s October 1 release says more than 90 participating banks and credit unions have a way to conduct bank-to-bank transactions. It does not publish a list of active users, completed transactions, daily settlement volume or time saved against their existing rails. Access is a product reach claim. It is not evidence that all those institutions are moving funds today. Crypto.news’ launch report correctly identifies VersaBank as issuer and notes the divergence in product labels.
BND describes participation as voluntary and says the product is for financial institutions only. This is not a coin a North Dakota resident can buy to speculate on the state economy. The public chain verifies transfers of a permissioned asset. It does not make membership open to every wallet. The bank’s use of Solana answers where transactions are recorded, while its own controls determine who may initiate them.
The number 90 is also different from the number of FDIC-insured institutions in North Dakota. BND’s ecosystem page lists 61 such institutions with $64.9 billion of assets as of the first quarter of 2026, and separately describes credit unions. The access count combines banks and credit unions, while the FDIC statistic counts an insured bank category. Dividing 90 by 61 to claim more than full statewide adoption would mix denominators and mistake eligibility for usage.
The dollar enters a designated account first
BND says minting occurs only after a transfer from a financial institution’s operating account to a designated account for the benefit of, or FBO, that institution is confirmed. The issuer can then create the token. Each participating institution maintains an FBO VersaBank Custody Account. The combined FBO account structure backs the tokens one-to-one with U.S. dollars, according to BND, with reconciliation through Fiserv and BND oversight.
Follow a hypothetical $1 million transfer. The sending bank moves $1 million of conventional funds into the designated FBO structure and, after confirmation, $1 million of tokens may be minted. A token sent to a receiving bank’s wallet represents movement within the scheme. Under BND’s described autoburn configuration, arrival at the receiving wallet triggers a burn instruction, which initiates the offchain leg. A burn reduces the circulating token supply; it does not by itself deposit spendable dollars in a receiving bank’s account. Account postings and reconciliation must complete the claim.
That separation is crucial. An onchain observer could see a million tokens minted and burned, but cannot inspect the FBO ledger or verify the timing of fiat credits from Solana data alone. A bank examiner can inspect both. A public user should not be asked to infer full dollar backing from a zero or low end-of-day token balance, because autoburn intentionally keeps the onchain balance low. The relevant audit trail joins mint confirmation, transfer, burn, reserve debit, receiving credit and any exception.
The BND page says funds movements are netted daily against an FBO VersaBank Concentration Account held at BND. It describes BND and Fiserv pushing and pulling ACH files among FBO accounts. That means a near-instant token event and a later conventional banking process can coexist in one transaction. The public chain’s finality is one component, not an assertion that every offchain ledger, ACH entry and customer balance is final in the same instant.
The bearer of the claim is the uncomfortable question
Fiserv says VersaBank issues the coin and manages reserve assets. BND says each institution maintains an FBO VersaBank Custody Account, with daily netting against the concentration account. Those descriptions point toward the issuer and account structure, but they do not publish every participant contract, insolvency provision or redemption agreement. One should not invent a blanket state guarantee because the state bank sponsors the project.
A participating bank should ask which legal entity owes it the dollar at each stage. Before minting, it holds a conventional account claim. While the token is outstanding, it holds whatever rights the program agreement gives a token holder against the issuer and underlying account structure. After autoburn, it must rely on the promised account settlement. The exact transition of rights is a matter of documentation. It is not settled by the word stablecoin or deposit in a press release.
This is an especially useful question when a transfer straddles the end of the banking day. If the token burned at 11:58 p.m. but an ACH file or concentration-account adjustment posts later, where is the recipient’s legally enforceable value during that interval? The program may have procedures that cover it. BND’s public page does not set out a complete legal waterfall. An examiner and participating bank should be able to trace it from the participant agreement and books.
The alternative interpretation deserves space. BND describes this as a token deposit rail that keeps deposits and oversight within a regulated ecosystem. The account structure and institutional limitation support that goal. A banking network can offer a new way to move value without inviting retail circulation or a public secondary market. But the benefits are strongest when the sponsor discloses the participant’s claim and demonstrates that daily netting consistently matches token events.
A failed payment has several possible states
Consider an institution that submits a $500,000 transfer at 10 p.m. The instruction can fail before the token is minted because its FBO funding was not confirmed. It can fail after mint but before broadcast because a wallet policy rejects the destination. It can be broadcast yet not reach the receiving bank because the address is frozen or wrong. It can reach the wallet and burn while the receiving account credit waits for an offchain reconciliation. These are four different incidents, each with a different place to look for the dollars.
An operator should be able to answer, for each state, whether the sending bank still owns a conventional dollar claim, whether a token remains outstanding, and whether the recipient may already credit its customer. A dashboard that labels the whole instruction failed or complete without separating these states can hide a double payment or a trapped balance. The exception log needs a unique reference connecting the bank instruction, reserve movement, mint signature, chain transaction, burn and resulting account entry.
Suppose the first instruction times out and a clerk retries it. If the first token transfer actually succeeded, the retry could send another $500,000 unless the system recognizes the original instruction identifier. Conversely, if the retry is suppressed but the first instruction never reached the chain, the receiving bank may be waiting on a payment that will not arrive. Idempotency, the rule that repeating an instruction does not repeat its economic effect, is a routine payments concept with an unusually visible onchain consequence here.
The Fiserv and BND descriptions do not publish an exception-rate series or a full failure-state diagram. That is a gap in public evidence, not proof the operators lack controls. A bank joining the network should request test cases for each state and an incident escalation matrix naming the party empowered to fix it. The more automatic the burn and posting, the more important it is to know who can pause an inconsistent instruction without making a second inconsistent entry.
Permissioned token, permissionless ledger
BND says the asset uses Solana’s public chain with built-in access controls and cites freeze and clawback capabilities available through Token-2022 extensions. This architecture separates validation from eligibility. Solana validators process transactions according to network rules; the issuer’s token rules and participant onboarding decide which institutions can hold or move Roughrider Coin. Public visibility of a transfer does not mean public redemption.
The freeze function is operationally useful for suspected fraud or sanctions compliance. It is also a power that must be governed. Who can trigger it, how many approvals are needed, whether a mistaken freeze can be reversed, and how the receiving bank is notified are program questions. A clawback may help resolve an erroneous transfer, but it qualifies any shorthand claim that the token leg is absolutely irreversible. The precise rights depend on the deployed token configuration and the legal terms, not the generic capability of a Solana extension.
Fireblocks-secured multiparty computation wallets divide signing authority, according to BND. That design can reduce exposure of a single secret. A participating bank still needs to know who can approve a transfer, change the policy, recover access or disable a compromised endpoint. If Fiserv’s Commercial Center is the only interface used by staff, its authorization rules become part of the custody perimeter even though the transaction settles on Solana.
The same goes for network outages. A chain can be operational while the bank portal, issuer approval engine or Fireblocks service is unavailable. Conversely, an institution can prepare a payment in the portal while Solana cannot finalize it at the expected time. A service-level claim should specify the whole route, measured from bank instruction to spendable fiat at the recipient, and disclose exclusions. Chain throughput alone measures a narrower step.
The comparison with ACH has to use one clock
BND’s page contrasts near-instant, around-the-clock token settlement with legacy ACH and wire cutoffs. It lists approximate per-transaction costs in a table, including about one cent for Roughrider Coin, but these are sponsor estimates, not a public independent measurement of all-in bank costs. There may be platform fees, liquidity costs, compliance expense and exception handling outside the token transfer fee. A claim of savings requires the same endpoints and cost categories for each rail.
For an ordinary bank-to-bank transfer, the total elapsed time can be partitioned into instruction, approval, funding, token mint, chain confirmation, token burn, fiat posting and reconciliation. If a token crosses the chain in seconds but the recipient can use dollars only after an ACH cycle, the operational benefit is smaller than the settlement headline suggests. If the participant credits usable funds immediately under contractual rules while netting later, the risk has shifted to the institutions that advance that liquidity. Both models can be legitimate; their economics differ.
An apples-to-apples test would take a sample of matched payments, including nights and weekends, and measure the median and worst-case time from authorized instruction to available recipient balance. It would count manual exceptions and failed transactions, not only successes. It would also compare the intraday funding tied up in the FBO accounts with conventional rails. A faster message that requires more idle dollars may not be cheaper for every community bank.
Crypto.news’ earlier account of the original Roughrider plan predates this live platform. The launch is a real step beyond an announcement. It does not yet provide the public throughput and cost dataset needed to score the step against ACH, wires, FedNow or RTP across a year of operation.
Words such as bank, deposit and state-owned invite a reader to assume a familiar guarantee. Roughrider’s public description does not, by itself, identify the beneficiary of any federal deposit insurance on each underlying FBO balance, the conditions for pass-through coverage, or how an individual institution’s claim would be treated if an intermediary failed. Those are matters for account records, contracts and applicable insurance rules. The token’s branding cannot supply them.
The same caution applies to state backing. BND’s oversight role is substantial, but it is separate from VersaBank’s stated role as issuer and reserve manager. An institution should not treat sponsorship as an unconditional state promise unless the governing documents say so. If a reserve account is held at a bank, the risk analysis should identify that bank, its account title and any concentration of funds. If several institutions share an FBO concentration structure, their underlying allocations need a reliable subledger.
There is a practical way to test the proposition before a large payment. Have the participant’s legal and treasury teams trace one dollar through each entity and account, including an issuer failure, a platform outage and an erroneous transfer. At each point ask who owes whom, which record proves the amount, and which party can instruct a correction. If two parties both appear to owe the same dollar, investigate double counting. If neither clearly owes it between burn and fiat credit, investigate a gap. The answer may be sound, but the public announcement does not supply the complete contract chain.
Fiserv’s platform is bigger than this coin
Fiserv says its platform can support tokenized deposits, global currency accounts, cross-border payments, card issuance and treasury automation. Those are platform capabilities or prospective use cases, not evidence that Roughrider Coin is already a public retail payment network. Its first live case has a narrow geography and institutional membership. That narrowness may be a strength: the participants have established banking relationships, shared operational channels and a bounded reason to exchange value.
The broader bank stablecoin integration trend reported by crypto.news shows why payment software vendors want reusable rails. This transaction, however, has a specific issuer and a specific North Dakota account loop. General platform ambition cannot answer a participant’s redemption question.
There is a naming collision with FIUSD. Crypto.news reported Fiserv’s prior FIUSD plan, which discussed a separate bank-friendly stablecoin and Solana integration. Roughrider Coin is a specific BND use case issued by VersaBank. Treating the two as the same asset would erase the issuer and redemption structure that determine the holder’s rights.
Expansion creates a new question. A token useful for transfers among BND partner institutions may not work unchanged in a national, cross-border or consumer setting. Onboarding rules, reserve structures, sanctions controls and the redemption party would need to fit the new users. A platform can reuse software without reusing every legal promise. That is why the first deployment is a useful case study rather than a universal template.
The strongest argument for the project comes from the banks’ current constraints. Smaller institutions may struggle to build their own continuous settlement systems, and access through a familiar Fiserv interface could reduce that barrier. BND says participation is voluntary. The counterweight is dependence on a stack of providers. A shared platform may simplify each bank’s work while concentrating operational failures across the network. Both effects should show up in measured uptime, exceptions and recovery procedures.
The difference between a bank deposit claim and a token holder’s claim also matters across the industry. Crypto.news’ reporting on the bank debate over stablecoin rewards shows how sharply banks distinguish deposits from circulating tokens when assessing funding. Roughrider’s restricted membership and FBO account design deserve analysis on their own terms, not an assumption that every token has the same effect on deposits.
The reserve math must include tokens in transit
At a cutoff, a simple balance equation is outstanding token liabilities equals dollars assigned to backing, subject to the program’s exact legal definition and any disclosed settlement items. Autoburn makes the left side jump down quickly. The auditor cannot stop there. It must also count any bank entitlement arising from a token already burned but not yet credited in the receiving account. Otherwise a low token supply could coexist with a large unsettled obligation hidden outside the displayed reserve ratio.
Take a toy end-of-day example. Assume $4 million of tokens remain outstanding, $2 million have burned but await recipient account posting, and $6 million of funds are in designated backing and concentration accounts. The economic claims to reconcile may total $6 million, even though the onchain supply is only $4 million. This is an illustration of accounting scope, not a claim about actual Roughrider balances. If a reserve report compares $6 million with only $4 million, its 150% ratio could look exceptionally strong while saying nothing about the pending $2 million. If the report covers both claims, the ratio is 100% in this stylized case.
The reverse problem is also possible. Funds may have arrived in a designated account while minting is pending, making backing dollars temporarily larger than issued tokens. The extra balance is a liability to the funding institution even though it is not yet token supply. A useful assurance report specifies which deposits are awaiting mint, which burns await offchain credit and which amounts are truly unencumbered. It should disclose whether balances are measured continuously, at business-day close or at a sample time.
BND’s daily netting means a public chain observer cannot do that complete calculation. Participating banks and supervisors can inspect the account records, and an independent attestor could summarize them without revealing each payment. The program’s maturity will show in whether it publishes such a scope statement, not just a headline backing percentage.
Public verification stops at the ledger boundary
An outside observer can examine Solana transactions if the mint and relevant addresses are disclosed. That observer might calculate minted supply, burns and transfer counts. The observer cannot independently see every bank’s FBO balance or legal account entitlement. A reserve attestation, if published, should state its cutoff, account scope, liabilities covered and whether tokens in transit or awaiting autoburn were included. One-to-one backing is an assertion to reconcile, not a property bestowed by a public chain.
BND says the program follows a GENIUS Act framework including reserve backing, attestations, disclosures and compliance controls. The practical question is when and where these assurances are published and how frequently they match the live obligation. A bank participant will have contractual and supervisory access that the public does not. A reader should not infer absence of controls merely because full account data are private, nor infer verified reserves merely from a sponsor statement.
The first meaningful performance report would publish active institution count, transaction count and value, median end-to-end settlement time, failed and reversed attempts, and reserve or liability reconciliation exceptions for a specified period. None of those requires publishing a community bank’s confidential customer payments. Each would make the network’s claimed gains testable and distinguish enrollment from use.
The coin is live. Whether it becomes better banking infrastructure will depend on the boring links between a token event and a dollar claim. The more successful the network becomes, the more important it is to show those links clearly.
What to watch
- Active usage: Count institutions that actually transact, not just the more than 90 with access.
- End-to-end timing: Measure authorized instruction to usable recipient balance, including the offchain account leg.
- Reserve reconciliation: Look for dated attestations that tie token liabilities, FBO accounts and unsettled amounts together.
- Exception records: Check whether freezes, reversals, failed burns and outages are reported with resolution times.
- Legal claim: Read participant terms identifying the issuer’s redemption obligation and BND’s separate oversight role.
FAQ
Can anyone buy Roughrider Coin?
No. BND says it is a permissioned asset for financial institutions and is not available to retail consumers or investors.
Did Bank of North Dakota issue it?
BND sponsors and oversees the program. Fiserv and BND identify VersaBank USA as the issuer responsible for minting, burning, custody and reserves.
Why use Solana if only banks can participate?
The public network processes and verifies permitted token transactions. Eligibility and redemption remain controlled by the banking program.
Are more than 90 banks already using it?
Fiserv describes access for more than 90 banks and credit unions. It has not supplied a public active-user or transaction-volume count in its launch release.
What happens to a token when the receiving bank gets it?
BND describes an autoburn instruction on receipt, followed by offchain account settlement and daily netting. The bank records complete the dollar movement.
Does one token always equal one dollar?
The program says tokens are backed one-to-one. An independent assessment requires matching liabilities to reserves and examining contractual redemption rights at a dated cutoff.
Is this the same as Fiserv’s FIUSD?
No. Roughrider Coin is the BND-sponsored institutional use case issued by VersaBank. FIUSD was described as a separate Fiserv plan.
What should a participating bank verify first?
It should trace the dollar claim from funding through mint, transfer, burn and account posting, including exceptions and issuer failure. This is educational analysis, not investment advice.
Disclaimer: This article is for information and educational purposes only and does not constitute financial or investment advice. Figures reflect regulatory filings and reporting available at the time of writing and change with each disclosure. Nothing here is a recommendation to buy, sell, or hold any security or asset. Always do your own research. Information is accurate as of October 2, 2026.
