For 15 years, moving tokenized dollars meant renting space on someone else’s blockchain, and the landlords were generalists: chains built for everything, where payments queued alongside collectible mints and margin trading for the same block space. This year the payments industry stopped renting.
Tempo, the chain incubated by Stripe with the crypto investment firm Paradigm, reached mainnet March 18, according to CoinDesk’s launch coverage. Arc, Circle’s chain, launches Sept. 16. Both are purpose-built for one workload, stablecoin payments, and both arrive with partner lists that read like a payments-conference badge scan.
Banks are on those lists by design. Deutsche Bank and Standard Chartered helped specify Tempo, and midtier American institutions have joined since, because a chain whose governance answers to payment companies is a chain a bank examiner can be walked through. The generalist chains treated payments as one workload among many; the payment chains treat everything else as someone else’s workload.
The scoreboard at launch
For orientation: Tempo has been live since March 18 with its machine-payments protocol shipping at launch and an expanding roster of banks and platforms building on it. Arc arrives Sept. 16 with its scope published and its cohort unannounced.
Neither has disclosed production payment volume from a named partner, and Ethereum and Solana continue to carry the overwhelming majority of stablecoin activity while this plays out. That is the fact every claim about the payment chains should be measured against for at least another year.
Tempo’s six-month sprint
Tempo went from announcement to mainnet in roughly six months, a fintech product cycle rather than a blockchain roadmap. The design-partner list did most of the talking: Visa, Mastercard, Deutsche Bank, Standard Chartered, Revolut, Nubank, Shopify, DoorDash, Ramp, and, most discussed, OpenAI and Anthropic.
Post-launch additions include Coastal Community Bank, the financial platform ARQ, Fifth Third Bank, the remittance firm Felix and Howard Hughes Holdings. An advisory unit followed in April to shepherd corporate adoption, according to Fortune’s reporting, and standing up the consulting wrapper in month two says the pipeline’s blocker has already moved from awareness to implementation.
The list sorts into three constituencies. Payment incumbents are hedging the rail layer the way they hedged the token layer with Open USD: if flows move on-chain, be a design partner of the chain. Banks, including midtier American institutions, are positioning for stablecoin settlement without betting on a public generalist chain their examiners distrust. And the AI firms are the tell for the design brief.
The machine-payments bet
Tempo launched alongside a Machine Payments Protocol, an open standard for software agents to hold balances, negotiate and settle autonomously. That is a thesis about where payment volume comes from next, and the reasoning is worth spelling out.
An AI agent booking freight or settling per-call API usage cannot open a checking account, pass card-network onboarding or wait two days for ACH. It can hold a wallet and sign a transaction. List what agent money requires and token rails keep answering: identity that is cryptographic rather than documentary, spending authority with hard limits enforced by the rail, settlement finality in machine time, auditability by default, because the principal’s liability for an agent’s spending is the first question counsel asks.
The issuer side is selling the same thesis in its own words. Circle “launched the Agent Stack to put programmable money at the center of the agentic economy,” Jeremy Allaire, the company’s co-founder and chief executive, said in its second-quarter results, the corporate translation of the bet Tempo’s protocol makes in the open.
Whether agentic commerce becomes a meaningful payment share this decade is genuinely uncertain. The asymmetry is not: if machine-to-machine payments arrive, they arrive on token rails, because nothing else is machine-shaped. Tempo’s bet is to be standing where that lands, with OpenAI and Anthropic in the room while the standard is drafted.
The economics nobody has stated
One number is missing from Tempo’s public story: what it earns its owners. A chain incubated by a processor and a crypto fund, with no public token at launch, is an anomaly in an industry whose chains finance themselves by selling their own scarcity. The design-partner model implies the intended answer: Tempo as a cost center that makes its stakeholders’ higher-margin businesses cheaper and stickier, the way container standards enriched shippers rather than the standard’s authors.
Whether that restraint survives success is a real question. A neutral, tokenless utility co-owned by competitors is stable while volumes are small, and the history of shared payment infrastructure, card associations and clearinghouses alike, is a history of member utilities eventually corporatizing, monetizing and being litigated over. The chains have imported the industry’s oldest governance drama along with its participants.
Arc, from the other direction
Where Tempo is a processor’s chain, Arc is an issuer’s. Circle built it, Circle mints the settlement asset, and the launch scope covers privacy features for institutional flows, an agent stack, tokenized real-world asset support and a whitepaper for a native ARC token contemplated for governance and network operations, according to the company’s roadmap.
Arc completes Circle’s vertical: asset, settlement network, managed platform, now ground. It also sharpens the neutrality question that follows every issuer down the stack. USDC won by being everyone’s asset on neutral ground; on Arc, Circle is landlord, mintmaster and toll-setter at once, in the same season the card networks built Open USD on the argument that no supplier should hold that much. The first 90 days of third-party deployments will show whether the market reads Arc as public ground or company town.
What the engineering brief shares
Strip the branding and both chains optimize the same list, because the workload dictates it. Deterministic subsecond finality, since a checkout cannot probabilistically settle. Fees that are low, stable and predictably denominated, because a rail whose costs spike with a memecoin launch is not a rail. Native stablecoin primitives, including fee payment in the settlement asset so users never hold a volatile gas token. Compliance surfaces: travel-rule data, sanctions screening, auditability as a condition of institutions showing up. Throughput sized for card-network peaks.
Several generalist chains can approximate each item. The point of specialization is the bundle plus the governance: a chain whose operators answer to payments companies will keep the bundle stable in ways a general community cannot promise, and that promise is what banks are actually buying when they choose rails.
Read more: Circle turns USDC into a payments network with rails of its own • Visa opens a stablecoin mint for its bank clients
How institutions will arrive
Regulated adoption follows a sequence the industry can now recite. Observation first, with examiners and risk teams watching the rail run under someone else’s weight. Custody second, through a supervised custodian, the lowest-commitment touch. Settlement of existing obligations third, the step Visa’s multichain pilots normalized, where the chain substitutes for a wire without changing the business. Issuance and product last, because they create new exposure.
The payment chains compressed the observation phase by recruiting the observers as design partners. A bank that helped specify the rail has pre-answered half its own diligence, which is the deepest function of those partner lists: co-authored examination files wearing marketing clothes.
None of the partner banks has described its internal approvals in public, and none needs to for the pattern to show: institutions that spent 2024 observing from conference panels spent 2026 co-writing specifications.
The block-space economics underneath
One mechanism explains why purpose chains exist at all, and it is worth a paragraph of plain economics. On a generalist chain, payments bid for block space against every other use, so a speculative frenzy anywhere in the ecosystem raises the cost of a $4 coffee settlement everywhere. Payment businesses cannot price products against a fee that moves two orders of magnitude on someone else’s news.
The purpose chains fix the problem by construction: admit mostly payment workloads, denominate fees in the settlement asset and size capacity for peaks, and the fee curve flattens into something a chief financial officer can put in a model.
The same logic explains the compliance surfaces. On neutral ground, travel-rule data and sanctions screening bolt on at the application layer, unevenly. On an operator-governed chain they can sit in the protocol’s expectations, which converts a per-institution integration into a property of the ground itself. Institutions do not choose chains for ideology; they choose whichever ground makes their examiner’s checklist shortest.
What happens to the generalists
Stratification, not displacement. Ethereum remains the settlement layer of record for the token economy’s deepest liquidity, Solana built the consumer-payments performance case the new chains cite as table stakes and both appear on Visa’s issuance platform beside Tempo, which says the industry expects them to persist.
The realistic forecast: trading liquidity, DeFi collateral and composability stay where the ecosystems are, while institutional settlement, corridor flows and machine payments drift toward purpose chains that prize governance and compliance surfaces over composability. The generalists are already growing payment-chain features, isolated fee lanes, faster finality paths, enterprise tooling, and where the two curves cross determines the decade’s rail map.
Fragmentation is the cost either way: dollars on Tempo, Arc, Ethereum, Solana and Avalanche, with liquidity and tooling split across all of them. The industry’s working answer is pragmatic multichain, with Visa’s platform supporting three chains at launch and settlement pilots spanning nine. The likely equilibrium is a routing layer over many grounds, the way correspondent banking abstracts across currencies, with chains competing to be the best place for specific flows rather than for the whole pie.
What could still go wrong
Three failure modes deserve the file. Ecosystem thinness, because a payments chain without developers becomes a private network with extra steps, and both teams are spending heavily on tooling to avoid that fate. Governance capture, because the first time an operator-owner uses the ground to advantage its own product, every other participant reprices the neutrality promise. And regulatory reclassification, because a chain governed by identifiable payment companies is a far easier enforcement target than a diffuse protocol, which is a cost of the design as well as a comfort.
None of these is prohibitive. All of them are the price of trading the generalists’ chaos for operator control, and the purpose chains’ first crises will reveal what the trade actually bought.
The precedent worth remembering
In the 1970s, banks drowning in paper checks built the automated clearinghouses: purpose-built shared infrastructure, governed by users, eventually so successful it became invisible. The card networks began the same way, as bank associations running shared authorization rails, before public listings turned infrastructure into empire.
The payment chains are the third arrival of that pattern, with the founding tension intact: infrastructure built by competitors who need it neutral, governed by owners who would each prefer it were not. The ACH era resolved the tension with regulation and nonprofit governance. The card era resolved it with IPOs and antitrust litigation. The chain era has not chosen, and its two flagship launches embody the two candidate endings, Arc with a governance token, Tempo with tokenless restraint.
The historians will not need new theory for this one. The primary sources are the card networks’ own listing prospectuses and the ACH rulebooks, and both endings are already on the shelf.
Why the answer matters commercially
Every constituency on those partner lists is also a potential defector. A bank that settles on Tempo can settle on Arc the day the terms improve, and an issuer that mints on three chains can mint on five. Multichain operations, which the industry adopted to avoid lock-in, also strip the purpose chains of lock-in, so the grounds will compete forever on current terms rather than accumulated switching costs.
That is good news for users and a hard business for chain operators, and it is the single biggest difference between this infrastructure cycle and the card era it otherwise resembles.
Until a supervised institution discloses settlement over either chain, the honest status of both is the same: live or nearly live, superbly connected and unproven at exactly the thing they were built for, which is carrying other people’s money in volumes that make the governance questions worth fighting about.