PROTOCOL PROTOTYPE · REFERENCE MARKETSExplore reference markets →
NeuraMesh
Open finance app →

Compute without the gatekeepers

The Monopoly on AI compute ends today

AI compute shouldn’t belong to a few.
We’re building an open market where providers connect,
builders choose, and capacity moves on your terms.

Open finance app
Discover the open compute model
An open network. Not a walled garden.
One protocol. Five verifiable layers.
CAPACITY-CAPPEDIssuance bounded by declared pool supply
OBJECT-NATIVETransferable claims on Sui
PROOF-BOUNDWorkloads and receipts on Walrus
// System architecture

Built as infrastructure,
not a token wrapper.

The protocol separates market discovery from execution and verification. Each layer has one responsibility, an explicit trust boundary, and a durable on-chain or content-addressed record.

01
Discovery planeSearch · quote · route
Indexes capacity pools by hardware, region, delivery window, attestation state, and workload compatibility.
MARKET API
02
Claim layerIssue · split · transfer
Creates bounded ComputeClaim objects against a provider CapacityPool without mixing execution logic into market ownership.
SUI MOVE
03
Execution planeSchedule · run · isolate
The node agent pulls an authorized workload, allocates a matching runtime, and emits a reservation-bound execution receipt.
NODE AGENT
04
Evidence planeAttest · meter · challenge
Hardware capability, metering evidence, input commitments, and result references remain independently inspectable.
WALRUS
05
Settlement planeRedeem · settle · release
Successful delivery settles the reservation and releases outstanding pool capacity; expiry and disputes follow separate paths.
SUI + ORACLE
// Marketplace

Compute contracts,
made comparable.

Browse standardized reference markets by the attributes that actually change delivery: accelerator, memory, region, window, settlement unit, and attestation status.

Explore surface

Find the right compute before your workload starts.

The marketplace is designed for AI teams today and machine buyers tomorrow. Every market exposes the same contract schema so an agent can compare and route without bespoke provider integrations.

Open marketplace →
H100 SXM · 80 GBUS-EAST · MONTHLY DELIVERY
INSPECT →
A100 SXM · 80 GBEU-WEST · RESERVED INFERENCE
INSPECT →
L40S · 48 GBAP-SOUTH · LOW-LATENCY INFERENCE
INSPECT →
// Market thesis

A machine-native economy could make compute capacity easier to specify, exchange, and settle through programmable infrastructure.

Our interpretation of BlackRock Digital Assets Research, The Machine-Native Economy. The paper is directional research—not an endorsement of NeuraMesh, proof of demand, or investment advice.