Capacity Allocation Marketplace

Mechanism

The allocation mechanism is structured as a sequence of credentialed bilateral exchanges rather than as an auction or central-clearing process. A capacity-holder publishes a capacity declaration: a structured object identifying the holder, the location and physical scope of the capacity, the time window over which the capacity is available, the quantity and unit of measure, and the commodity class to which the capacity belongs. The declaration is signed under the holder's jurisdictional credential and admitted to the governance chain.

A capacity-user issues a request referencing one or more declarations. The request specifies the user's identity and credential chain, the requested quantity, the requested time subwindow within the declaration's window, and the priority tier under which the request is being made. Tiers are governance-declared and class-specific; a higher tier may, by rule, displace a lower tier under specified conditions. The request is itself a chain object and carries the lineage to support later replay.

Settlement is bilateral and pair-wise. The capacity-holder and the capacity-user produce a joint settlement record incorporating the declaration identifier, the request identifier, the agreed quantity and subwindow, the priority tier, the price terms (where applicable), and the fallback semantics that govern what happens if the holder cannot deliver. Each party signs the settlement under its own jurisdictional credential, and the settlement is admitted to the chain. Where the underlying infrastructure is shared among multiple holders, a coordination authority — itself credentialed under the relevant jurisdiction — co-signs to attest that the bilateral grant is consistent with the shared-infrastructure schedule. No central matching engine is consulted; the chain itself is the consistency mechanism.

Allocation rules — the structured definitions of priority tiers, displacement conditions, fallback semantics, and tier-specific eligibility — are credentialed objects in the chain, signed by the jurisdictional authority that promulgates them. A settlement references the rule by signed identifier; the rule's content at the moment of signing is fixed in lineage. If the authority later updates the rule, the new rule enters the chain as a successor object with its own identifier; pre-existing settlements continue to reference the rule version that was operative at signing, and any replay or audit applies the historical rule rather than the current one.

Admissibility of a settlement to the chain is conditioned on a multi-credential predicate evaluated at the moment of submission. The predicate verifies that the holder's signing credential resolves to an authority whose jurisdictional scope encompasses the declared physical location, that the user's signing credential resolves to an authority whose membership relation includes the requested commodity class at the requested tier, that the rule reference resolves to a rule object whose effective window encloses the settlement timestamp, and that the coordination authority's co-signature — where required — is attached and verifiable. A failure of any clause in the predicate causes the settlement to be rejected at admission rather than admitted and later disputed; the rejection is itself recorded as a chain event so that a holder cannot quietly resubmit a malformed settlement under different terms.

The mechanism explicitly accommodates partial settlements. A request for a quantity exceeding a single declaration's available capacity may be satisfied by a sequence of bilateral settlements against multiple holders, each settlement carrying the same request identifier and a settlement-sequence index. Partial-settlement composition is itself credentialed: the user's request includes a partial-acceptance flag that the holders rely on when admitting to a fractional share, and the chain enforces non-overlap so that the same physical capacity is not double-granted across the settlement sequence. When a request cannot be fully satisfied, the unsatisfied residual is recorded as a chain object so that downstream auditors can distinguish unmet demand from unrequested capacity.

Operating Parameters

Time-window granularity is parameterized per commodity class. For grid-balancing capacity, windows may be sub-second; for transportation capacity, minutes to hours; for data-center capacity, minutes to days. The granularity parameter governs the smallest indivisible unit of slot grant and determines the resolution at which fallback and replay operate. Granularity is declared by the jurisdictional authority for each class and bound to the rule object referenced by settlements.

Priority tiers are operationally ordered and class-scoped. Tier definitions specify, for each tier, the eligibility predicate (which credential classes may request at that tier), the displacement rule (whether and under what conditions the tier may displace lower tiers already settled), and the compensation owed to displaced parties. The compensation discipline is integral to the tier definition; an emergency tier that can preempt commercial slots, for example, will typically carry a structured compensation obligation to the displaced commercial party that follows from the same chain mechanism.

Fallback semantics handle the case in which a holder cannot deliver granted capacity. Parameters include the maximum permissible substitution scope (whether substitute capacity from the same holder, from a coordinated pool, or from any pool may be used), the notification timing required, the chain-recorded fallback record format, and the financial settlement that follows. Fallback is not implicit; it must be referenced by the settlement at signing time, drawn from a governance-declared catalog of fallback strategies, and itself credentialed.

Slot-grant durability — how long after the settled time window the grant remains lineage-retained for audit — is parameterized. Short-window classes may require retention measured in months; classes subject to long-tail regulatory inquiry may require retention measured in years or decades. Retention is bound to the class declaration and propagates with the grant.

Alternative Embodiments

In one embodiment, capacity declarations are pre-published in batches by holders and indexed by a directory peer that any user may query; the directory peer is not authoritative and does not match — it merely accelerates discovery. In a second embodiment, capacity declarations are issued reactively in response to user-published intent, with holders selecting which intents to respond to. In a third embodiment, both modes coexist: routine capacity is pre-published while specialized or peak capacity is issued reactively.

Coordination-authority embodiments differ. In one embodiment a single coordination authority co-signs all settlements within a jurisdiction. In a second embodiment, coordination authority is sharded by infrastructure subdomain, with each shard signing within its scope and the chain enforcing non-overlap structurally. In a third embodiment, coordination is itself bilateral between adjacent holders, with no separate coordination authority; consistency is then maintained by chain-level dispute mechanisms rather than by a co-signing party.

Pricing embodiments include fixed tariff (price terms determined entirely by published authority schedule), bilateral negotiated price (terms set by the holder/user pair within authority-declared bounds), and market-discovered price (terms derived from a publication of recent settlements within a class and window). In all pricing embodiments, the chain records the terms applied and the rule version that authorized them; the chain itself is price-neutral and merely retains the credentialed record.

Composition with commodity-class-plurality permits a single settlement to span multiple commodity classes — for example, a single transaction that grants both compute capacity and the network capacity required to deliver outputs from that compute. Each class contributes its own declaration, its own rule reference, and its own fallback semantics; the settlement is consistent only if all class-specific predicates are simultaneously satisfied at signing.

Composition With the Broader Architecture

Capacity allocation composes with the governance chain by construction: every declaration, request, settlement, rule, and fallback record is a chain object subject to admissibility, weighting, lineage retention, dispute, and replay. Composition with commodity-class-plurality permits multi-class settlements as described above. Composition with the dispute mechanism permits any party — holder, user, coordination authority, or third-party auditor — to raise a dispute against a specific settlement on grounds of rule misapplication, fallback non-compliance, or credential invalidity at the time of signing.

Composition with spatial adaptation is structural in the inverse direction: a deployed adaptation may consume capacity grants as one of its operational inputs, with the chain ensuring that the adaptation's runtime activity is bounded by the slots it has been granted. Composition with cross-jurisdictional operation permits settlements to span multiple jurisdictions provided each jurisdictional authority is represented in the credential chain of the settlement and the rule references include the cross-jurisdictional reconciliation policy.

Composition with the lineage-retention discipline ensures that every settled grant carries forward the complete decision context — the declaration text, the request text, the rule version, the coordination co-signature, and the fallback strategy — for the duration declared by the class retention parameter. Composition with the credential-revocation mechanism is forward-looking: revocation of a holder's, user's, or coordination authority's credential after settlement does not automatically void the settlement, because the rule version in force at signing is the binding instrument; instead, revocation triggers a structured review track in which dispute filings against settlements that depended on the revoked credential are evaluated against the rule's revocation-handling clause. This separation of revocation from automatic invalidation is essential for stability: a single late-discovered credential compromise cannot retroactively unwind a chain of dependent operations without explicit governance action.

Distinction From Prior Art

Wholesale capacity markets in electricity, transportation, and telecommunications generally rely on a central matching authority — an ISO, a slot coordinator, a clearing exchange — that is the structural source of allocation truth. Audit, dispute, and replay are properties of the central authority's records. Cloud-platform capacity reservations are similarly centered on the platform operator's reservation system, with terms enforced and settled inside the platform's own ledger.

The mechanism described here has no central matching authority. Settlements are bilateral; coordination is credentialed and may be absent, sharded, or peer-to-peer; and the rule under which any settlement was reached is itself a chain object whose historical content is fixed at the time of reference. Audit and replay are properties of the chain, not of any operator. Multi-jurisdiction composition is structural rather than the result of bilateral inter-operator agreements. The combination of credentialed jurisdictional rule-publication, bilateral pair-settlement, structured priority/fallback semantics, and chain-resident replay distinguishes the disclosed mechanism from prior allocation systems.

Disclosure Scope

This disclosure covers the capacity declaration, request, and bilateral settlement objects; the credentialed allocation-rule object and its versioning discipline; the time-window, priority-tier, and fallback parameter sets; the coordination-authority role under shared infrastructure; and the composition of capacity allocation with commodity-class-plurality, governance-chain dispute and replay, spatial adaptation, and cross-jurisdictional operation. Alternative embodiments — pre-published versus reactive declarations, single versus sharded versus peer coordination, fixed versus negotiated versus discovered pricing — are within scope. Implementations centered on a central matching engine, or in which allocation rules are not chain-credentialed objects, fall outside the disclosed mechanism.