LUXDocumentation
Docs/A-Chain (AI)

A-Chain (AI)

AI compute task management with TEE attestation (SGX, SEV-SNP, TDX, nvtrust). Provider registration, task assignment, epoch-based rewards, and Merkle anchoring to Q-Chain.

Overview

The AI Chain manages decentralized AI compute workloads with hardware-level trust through Trusted Execution Environment attestation. Supported TEE platforms include Intel SGX, AMD SEV-SNP, Intel TDX, and NVIDIA nvtrust.

Compute providers register on the A-Chain with their hardware capabilities and TEE attestation certificates. Tasks are assigned based on requirements, and results are verified through attestation proofs anchored to the Q-Chain via Merkle trees.

Epoch-based rewards incentivize providers to maintain high availability and performance. The A-Chain enables trustless AI inference and training without revealing model weights or input data.

Technical Specifications

TEE SupportSGX, SEV-SNP, TDX, nvtrust
Task TypesInference, Training, Fine-tune
AttestationHardware-level
RewardsEpoch-based
AnchoringMerkle → Q-Chain
Chain TypeOptional

Features

TEE attestation (SGX, SEV-SNP, TDX, nvtrust)
Compute provider registration
Task assignment and scheduling
Epoch-based reward distribution
Merkle proof anchoring to Q-Chain
Confidential AI inference

Quick Start

Install the Lux SDK and start interacting with AI Chain.

Installation

npm install luxnet

Example

import { AiClient } from 'luxnet/ai'

const ai = new AiClient('https://ai.lux.network')

// Submit an inference task
const task = await ai.submitTask({
  model: 'llama-3-70b',
  input: { prompt: 'Explain Lux Network' },
  tee: 'sgx',  // require SGX attestation
})

const result = await ai.getResult(task.id)

SDKs & Libraries

API Reference

Full API reference documentation is available at docs.lux.network.

View Full API Docs