Ledger compute like money, because it is
Meter GPU-hours as a first-class asset, convert delivered compute into invoices and settlement, and keep operators, tenants, and rails in one auditable record, at inference-provider volume.
The compute economy runs on IOUs nobody can audit
Billions in GPU spend, settled on spreadsheets and screenshots.
Compute is delivered in hours, billed in dollars, and settled in USDC and fiat, across operators, tenants, and financing partners. Usage lives in one system, invoices in another, and the underwriters funding nine-figure GPU deals have no book to underwrite against.
Delivered compute, billed usage, and collected cash live in three unconnected systems, and the mirror database drifts from all of them.
USDC and fiat settlement with unlocked pricing makes every conversion a bespoke booking.
Operator payouts and tenant invoices are computed in batch, disputed by email.
Unit economics shift with every hardware generation and workload mix, and financing partners underwriting GPU deployments have no auditable usage-to-revenue record.
_PROOF/ An AI compute payments platform runs its system of record on the open-source Formance ledger, from metered GPU-hours to settled fiat.
From GPU-hour to settled dollar
Delivered, billed, collected, and paid out: one chain, two units.
Delivered compute posts as a metered asset, invoicing converts it into a tenant payable and an operator receivable, collection settles in USDC or fiat, and the operator is paid: every step double-entry, both units reconciled.
The primitives the compute economy needs
Built on the open-source Formance ledger.
Asset-agnostic metering, dual-unit accounting, and multi-rail settlement, composed for compute marketplaces.
Ledger GPU-hours, tokens, or any usage unit with the same double-entry guarantees as money: delivered, reserved, consumed, and billed states all explicit.
asset-agnostic by design
Convert metered usage into tenant payables and operator receivables in one atomic exchange: the usage ledger and the money ledger never drift, and unit margin per workload is a query, not a model.
dual-unit atomicity
Read the Numscript docsCollect and settle across USDC and fiat rails with per-rail clearing models (gross with fee skim, or net on funding batch), each netting to zero by construction.
per-rail clearing
Hand financing partners an auditable usage-to-revenue history per operator and per deployment: the book that GPU lending has been missing.
a book to underwrite
Compute-economy grade
Usage converts to revenue atomically
GPU-hours out, dollars owed: one transaction.
One Numscript transaction bills delivered compute: the consumed GPU-hours move to billed state while the tenant's payable and the operator's receivable open in the same operation.
Trusted by builders
“Formance enabled us to ship new lending products faster by providing a customizable foundation that let us reliably express the complexities of our flow of funds.”
Embedded finance · reconciliation across 14 countries
See customer story
“Formance is the foundation of our Financial OS, the open-source approach lets us retain control over this key component.”
Healthcare · Financial OS
“Formance helped us kick off our move into fintech, with robust infrastructure and intuitive developer tooling.”
Benefits · fintech infrastructure
Built for regulated money.
Enterprise controls, the certifications auditors ask for, and an immutable record, out of the box.
09:41:07Z AUDIT gateway POST /api/ledger/v2/main/transactions 200 sub:ops@acme.io
09:41:09Z LOG id:4093 NEW_TRANSACTION ledger:main
09:41:12Z LOG id:4094 SET_METADATA ledger:main
09:41:15Z LOG id:4095 REVERTED_TRANSACTION ledger:main
09:41:18Z AUDIT gateway GET /api/ledger/v2/main/logs 200 sub:audit@acme.io
Keep going
Everything you need to evaluate and build on Formance.
Compute marketplaces, answered
01 / CAN THE LEDGER REALLY HOLD GPU-HOURS?
Yes: the ledger is asset-agnostic: define GPUH (or tokens, or credits) as an asset and it gets the same atomicity, precision, and auditability as USD.
02 / HOW DO USDC AND FIAT COEXIST?
Both post to the same model. Conversions are explicit transactions with rates and spreads booked, and each rail's clearing account nets to zero by construction.
03 / WHAT ABOUT VOLATILE PRICING?
Rates are captured per conversion at booking time, so the usage-to-revenue history stays exact even when pricing moves between delivery and settlement.
04 / WHY DOES UNDERWRITING CARE?
GPU financings are nine-figure bets on future usage revenue. An immutable usage-to-cash record per operator is the underwriting book that has not existed.
The system of record for the compute economy
See how Formance fits your marketplace: book a demo or start building.