No score assigned. Tempo is a Tier 2 monitoring target. This brief is qualitative analysis, not a rating. Linkmerica has published ACR scores for five Tier 1 agentic payment protocols; Tempo is not among them. An assessment may follow as coverage extends.
Visa operates a production validator node on Tempo — a materially different commitment from joining a consortium. The validator set is a permissioned institutional group, not a decentralised network. Where the Tier 1 cohort raised questions about undisclosed or unbuilt controls, Tempo's institutional backing is unusually concrete, and that concreteness is itself the custody question.
Tempo MPP: Permissioned Validators and the Session-Escrow Custody Surface
The Linkmerica Research Team July 30, 2026
1. Executive Summary
Tempo is the first Tier 2 protocol Linkmerica has covered in depth. Its distinguishing characteristic is the inverse of what the Tier 1 cohort showed: where those protocols raised questions about undisclosed or unbuilt controls, Tempo's institutional backing is unusually concrete—and that concreteness is itself the custody question.
Machine Payments Protocol (MPP), co-authored by Stripe and Tempo, is a purpose-built Layer 1 blockchain that launched on mainnet March 18, 2026. It processes payment flows for Stripe's stated $1.9 trillion annual volume base and integrated over 100 service providers at launch, including Alchemy, Dune Analytics, Shopify, Anthropic, OpenAI, and Merit Systems. Visa joined as an anchor validator in April 2026 and operates a production validator node following a six-month engineering integration.
The validator set is a permissioned institutional consortium, not a decentralised network. Named participants include Visa and Stripe-affiliated entities. A March 2026 analysis characterised it as a "permissioned payment network wearing blockchain aesthetics." This brief examines the custody implications of institutional consensus operation, session-based delegation, off-chain voucher settlement, and multi-rail liquidity. No ACR score has been assigned; an assessment may follow as the framework's coverage extends.
2. What MPP Is
Machine Payments Protocol is an open standard designed for sub-100ms latency micropayments between machines and autonomous agents. The core primitive is session-based: a payer and payee initialise a session in approximately 500ms, after which cumulative payments are authorised off-chain via EIP-712 signed vouchers and settled in batches on-chain.
Session initialisation establishes a temporary escrow envelope. The payer delegates authorisation to a session key using EIP-712 signatures, enabling the autonomous agent or service provider to increment payment vouchers without requiring main wallet interaction for each transaction. Vouchers accumulate off-chain; settlement happens asynchronously in batches, collapsing hundreds or thousands of micro-authorisations into a single on-chain state change.
MPP supports multi-rail settlement: stablecoins, card payments via Visa, and Bitcoin Lightning. Visa extended MPP to card payments across its network following its April 2026 validator integration. Lightspark extended MPP to Bitcoin Lightning, enabling cross-network payment routing through the same session primitive.
Safe protocol multi-sig integration provides 2-of-3 and 3-of-5 threshold controls for custodians managing MPP sessions on behalf of institutional principals. This layer, which cites MiCA compliance in its custody product documentation, allows institutions to impose signing thresholds over session key delegation and voucher settlement approval.
The design prioritises latency reduction and API simplicity. Traditional blockchain payment flows require per-transaction wallet confirmation; MPP's session model collapses that interaction into a single upfront authorisation, enabling machine-to-machine streaming payments that match the speed requirements of real-time compute, API metering, and autonomous agent workflows.
The architecture treats sessions as temporary payment envelopes. Public materials describe initialisation, voucher accumulation, and batch settlement, but Linkmerica has not identified published documentation of session expiry, cap, or termination mechanics. The three custody surfaces the model implies are session initialisation (delegation), voucher accumulation (off-chain authorisation), and batch settlement (on-chain finality).
3. The Permissioned Validator Question
Visa operating a production validator node is a materially different fact from Visa joining a foundation. Institutional operation of consensus infrastructure concentrates trust in named, regulated entities. The validator set is a permissioned consortium; participation is not open, and block production is governed by identified institutional participants including Visa and Stripe-affiliated entities.
This reduces some risks while creating others. On the risk-reduction side: accountability is explicit, legal recourse is available against known corporate entities, and regulatory reachability is direct. Permissioned validators are subject to existing financial services compliance regimes, internal controls frameworks, and audit obligations that decentralised validator sets generally are not. For an institutional custodian evaluating counterparty risk, the ability to subpoena Visa or Stripe in the event of a dispute is not theoretical.
On the risk-creation side: censorship capability is concentrated, coordinated failure becomes possible, and gatekeeping is inherent. A permissioned validator set can, by design, refuse to process certain transactions, blacklist certain addresses, or comply with government orders to freeze or reverse state. The March 2026 analysis that characterised Tempo as a "permissioned payment network wearing blockchain aesthetics" is addressing this architectural reality: consensus is controlled by institutions whose business models and regulatory obligations may conflict with the transaction neutrality typically associated with public blockchains.
This is not automatically worse for institutional custody—it is different, and the difference must be assessed rather than assumed. Linkmerica's scoring of Visa TAP at 7.2—the highest risk of the five Tier 1 protocols—reflects a centralised key directory operated solely by Visa, with a federated model specified but unimplemented. TAP places trust in Visa at the identity layer. MPP places trust in Visa (and other validators) at the consensus layer. Same firm, different layer, different custody exposure.
For a corporate treasury or fund custodian, the question is whether permissioned consensus operated by regulated payment giants creates more custody confidence or less than a decentralised validator set with pseudonymous operators. The answer depends on which risks the institution prioritises: legal recourse versus censorship resistance, operational transparency versus network neutrality, regulatory compliance versus protocol immutability.
The architectural choice is explicit. Tempo and Stripe designed MPP as an institutional payment rail, not as a public utility. The validator set reflects that design. Custody risk assessment must proceed from that premise, not from the expectations set by proof-of-stake public chains.
4. Session Escrow and the Off-Chain Voucher Surface
Off-chain vouchers are where custody risk concentrates. In MPP's session model, value is authorised off-chain and settled in batches. The window between authorisation and settlement is a distinct exposure. During that window, vouchers exist as signed but unbroadcast commitments—enforceable in principle, but not yet reflected in on-chain state.
The IETF specification for MPP assumes an attacker with network observation capability. This is standard threat modelling for payment protocols, but it exposes two specific vectors: replay and forgery. A replay attack would involve resubmitting a previously valid voucher. A forgery attack would involve generating a fraudulent voucher using a compromised or improperly delegated session key. Linkmerica has not reviewed the specification's countermeasures against either vector and does not assert their presence or absence.
Delegation via EIP-712 signatures creates the convenience that makes sub-100ms latency possible, but it also moves authorisation off-chain and into the session key's control. If a session key is compromised, an attacker may be able to generate valid vouchers for as long as the session remains open. Linkmerica has not identified published documentation of session duration limits, cumulative caps, or revocation mechanics, which means the bounds of that exposure cannot be determined from public sources.
Institutions evaluating MPP custody should interrogate four operational controls: voucher expiry enforcement (how quickly does the protocol invalidate stale vouchers?), session revocation mechanisms (can a session be emergency-closed mid-stream?), nonce and idempotency guarantees (what prevents replay at the protocol level?), and session key storage practices (where do delegated keys reside, and under what access controls?).
The Safe protocol multi-sig integration addresses part of this. By wrapping session delegation behind 2-of-3 or 3-of-5 signing thresholds, institutions can impose human-in-the-loop approval over session creation and settlement. Whether those thresholds apply at session creation, at settlement, or at per-voucher granularity is not documented in materials Linkmerica has reviewed. That distinction determines how much autonomous authority a delegated session key holds once a session is active, and it is a question institutions should put directly to Tempo.
Off-chain vouchers are not inherently fragile. The underlying trade-off is speed against containment: the same batching that delivers sub-100ms latency is what creates a window between authorisation and settlement. Whether that window can be bounded — and how tightly — is the question an institution must answer before treating MPP sessions as custody-safe, and it is not answerable from public materials today.
5. Multi-Rail Settlement and Liquidity Fragmentation
MPP supports three settlement rails: stablecoins, card payments via Visa, and Bitcoin Lightning through Lightspark's extension. Each rail has distinct regulatory and operational profiles, and each introduces different liquidity and finality assumptions.
Stablecoins on MPP inherit the reserve and compliance characteristics of their issuers. Tempo's positioning aligns with GENIUS Act requirements: 1:1 reserve backing, monthly public reporting, and compliance programmes. Built-in privacy measures and user-level permissions provide compliance hooks. However, no explicit regulatory approval or licence is disclosed in available materials. Alignment claims exist; disclosed regulatory status does not.
Card settlement via Visa routes through existing card network infrastructure. Visa's April 2026 integration extended MPP to card payments across its network, meaning an MPP voucher may settle through card rails. Card settlement operates under different finality assumptions from on-chain settlement, and institutions should establish with Tempo and Visa which dispute, chargeback, and liability rules apply to agent-initiated card transactions under MPP. For institutions accustomed to blockchain's cryptographic finality, card rail settlement represents a regression to legacy payment finality assumptions.
Bitcoin Lightning settlement via Lightspark adds a third finality model: Lightning channel state, which is off-chain until channel closure forces on-chain settlement to Bitcoin's base layer. Lightning introduces channel liquidity constraints, routing failures, and watchtower requirements. It also introduces Bitcoin's regulatory ambiguity in jurisdictions where Lightning's legal classification remains contested.
The purpose-built Layer 1 architecture creates an isolated liquidity pool relative to established chain ecosystems. This is liquidity fragmentation risk: value settled on Tempo is not automatically fungible with liquidity on other networks. MPP's use of EIP-712 signatures indicates Ethereum-standard signing, but Linkmerica has not verified the extent of EVM compatibility or what bridging paths exist. Institutions should establish both before assuming liquidity portability.
For institutions, multi-rail settlement offers optionality but requires rail-specific operational readiness. Card settlement demands card network compliance infrastructure. Lightning demands channel management and watchtower monitoring. Stablecoins demand reserve transparency and redemption path clarity. An institution cannot custody MPP sessions passively; it must custody across three rails with three operational postures.
6. Where Tempo Sits Against the Scored Cohort
Linkmerica has published ACR scores for five Tier 1 agent payment protocols. No score has been assigned to Tempo. The following is qualitative positioning, not a rating.
The scored cohort exhibits five failure modes. Architecture: Visa TAP at 7.2, the highest risk of the five—a sole directory of agent public keys, with the federated model specified in TAP's own documentation left unimplemented. Opacity: Mastercard AP4M at 6.6—controls not publicly disclosed, a due-diligence gap rather than a demonstrated weakness. Documented risk: Google AP2 at 6.4—published research demonstrated prompt injection producing cryptographically valid mandates. Immaturity: Ripple XRPL at 6.2—Phase 1 status, 1.4 million transactions against Base's 119 million. Delegation: Coinbase AgentKit at 6.0, the lowest risk of the five—its own documentation warns against the public facilitator for production mainnet EVM routes.
Tempo does not cleanly map to any of these. It is not opaque—Visa's validator role is disclosed, the session model is specified in an open IETF standard, and Safe multi-sig integration is documented. It does not exhibit documented exploit risk in the same way AP2 does. It is not immature—over 100 integrations at launch and Stripe's $1.9 trillion volume base represent production scale. It does not rely on a third-party facilitator in the way AgentKit's documentation cautions against.
Tempo's custody profile most closely resembles Visa TAP's centralisation, but at a different layer. TAP centralises identity; MPP centralises consensus. Both concentrate trust in named institutions. The March 2026 "blockchain aesthetics" critique applies to both: the use of blockchain infrastructure does not, by itself, decentralise risk if the validator set or key infrastructure is controlled by a small number of institutional operators.
Tempo may represent something the existing taxonomy does not yet cover: concentrated institutional operation as a deliberate design choice, not as an interim state or oversight. The permissioned validator model is not a gap to be filled by future decentralisation—it is the architecture. Custody assessment cannot proceed by comparing Tempo to decentralised networks, because it is not one and does not claim to be.
The session-escrow surface—off-chain vouchers accumulating between settlements—adds a temporal exposure dimension that Linkmerica did not encounter in the same form when scoring the five Tier 1 protocols. MPP batches authorisations across time within a session envelope rather than settling each authorisation as a discrete event. The custody question is not whether sessions are secure in principle, but whether institutions can operationalise session hygiene—expiry, revocation, cap enforcement, key storage—at scale.
7. Monitoring Position
Tempo is a Tier 2 target in Linkmerica's monitoring surface, swept weekly with major releases flagged. No ACR score has been assigned to Tempo. An assessment may follow as the framework's coverage extends to purpose-built payment chains and session-based settlement primitives.
The regulatory gap is material. Tempo's positioning aligns with GENIUS Act and MiCA compliance requirements, and the Safe-based custody product cites MiCA compliance explicitly. However, no regulatory approval, licence, or supervisory relationship is disclosed in available materials. Alignment is not authorisation. Institutions relying on Tempo for custody workflows should independently verify regulatory status in their operating jurisdictions.
Visa's validator role is the most significant institutional signal. A six-month engineering integration resulting in a production validator node represents capital and operational commitment at a scale that advisory board membership or pilot participation does not. Mastercard's AP4M framework recognised MPP alongside x402 as a standardising protocol for agent payments in June 2026, indicating cross-network institutional validation.
Linkmerica will monitor three areas: validator set composition and governance (whether additional institutions join, and whether governance terms are disclosed), session-escrow operational incidents (replay, forgery, or compromise events in production), and regulatory status changes (licence grants, supervisory guidance, or enforcement actions). Material updates will be reflected in subsequent briefs.
The $500 million raise reported in connection with Tempo and Paradigm, if confirmed, would represent institutional capital commitment consistent with the validator profile. Linkmerica does not score funding as a risk input, but it is a signal of staying power and operational capacity.
Tempo's custody risk cannot be assessed in isolation from its institutional operator profile. The permissioned validator set is not a bug—it is the design. Custody assessment must proceed from that premise and evaluate whether concentrated institutional operation, session-based delegation, off-chain voucher accumulation, and multi-rail settlement create acceptable risk within the institution's operational tolerances and regulatory constraints.
Disclosure: Anthropic is listed among Tempo's launch integrations. Linkmerica's research and scoring pipeline uses Anthropic's API as tooling. Anthropic is not the subject of this brief and has no involvement in Linkmerica's methodology, scoring, or conclusions. Linkmerica accepts no payment from any rated entity or from its tooling vendors.
Linkmerica is a trade name of CASPO LLC. LISR scores and research are for informational purposes only and do not constitute financial or investment advice. This brief is based on publicly available information as of July 30, 2026.