Noos will launch its Layer 1 mainnet at 00:00 UTC on September 5, 2026, moving the decentralized AI infrastructure project from testnet operations into its production phase. In its official mainnet announcement, Noos said the network enters launch with more than 3,500 distributed computing nodes, over 50 ecosystem partners and 5,000 Genesis agent identities. The transition is intended to turn the project’s existing AI-agent infrastructure into a live network for execution, verification and settlement.
Noos describes itself as an economic layer where AI agents and applications can discover computational skills, invoke them and settle payments on-chain. Its network website says Proof of Agentic Contribution, or PoAC, measures verified execution and distributes rewards based on completed work rather than simply hashing power or token ownership. The mainnet launch will test whether that contribution-based model can operate under real economic activity rather than incentivized testnet conditions.
PoAC Links AI Work to On-Chain Settlement
The network divides responsibilities between several participant types. Intelligent Compute Nodes provide execution capacity, validation nodes verify contributions, while agent-service infrastructure allows developers to publish callable capabilities that can be used by other agents and applications. Noos is attempting to connect AI execution with a settlement system in which verified work can generate automatic on-chain compensation.
That architecture reflects a broader challenge emerging around autonomous software. The U.S. National Institute of Standards and Technology launched its AI Agent Standards Initiative in 2026 specifically to promote secure, interoperable systems capable of acting autonomously on behalf of users. Noos approaches that problem from an economic angle, focusing on how independent agents can establish identities, prove contributions and exchange value through shared infrastructure.
Noos has also described zero-knowledge proofs as part of its longer-term verification architecture. The NIST definition of zero-knowledge proofs explains that the technique allows one party to demonstrate that a statement is true without revealing additional underlying information. Applied to decentralized AI, that type of cryptography could support verifiable computation while limiting disclosure of sensitive inputs, although Noos still needs to demonstrate how its implementation performs at production scale.
Mainnet Shifts Focus From Nodes to Usage
The project says its token model is structured around contribution-driven distribution rather than conventional private allocations, with its public materials describing zero pre-mine and rewards generated through network participation. Its roadmap places agent-to-agent settlement, task scheduling and broader AI workload activity within the mainnet phase. The economic test will therefore be whether actual demand for AI execution replaces testnet incentives as the principal driver of node participation.
Noos already claims more than 100,000 community members and thousands of distributed nodes, but those figures measure ecosystem preparation rather than production usage. After September 5, metrics such as completed agent tasks, settlement volume, active nodes and paid computational workloads will provide clearer evidence of adoption. The launch establishes Noos as a live AI-focused Layer 1, but sustained agent activity will determine whether its PoAC model develops into a functioning machine-to-machine economy.