Network Architecture

Built for deterministic settlement and sovereign AI throughput.

Aethelred is engineered as a dual plane system: a CometBFT financial settlement layer and an enclave-native inference mesh. Protocol economics reward useful compute while preserving predictable state transitions and governance controls. Pair this architecture with the developer documentation, the governance framework, and the token design that secures incentives.

Proof of Useful Work Post quantum security TEE attestations CometBFT finality
Consensus

Proof of Useful Work

Aethelred replaces hash-only mining with inference-backed validator work. Instead of burning energy on arbitrary puzzles, validators execute signed AI jobs in TEE environments and receive rewards for measurable utility.

Job lifecycle

1. Job commitment

Clients submit workload metadata, latency targets, jurisdiction constraints, and model policy requirements.

2. TEE scheduling

Orchestrator assigns workload to validators based on attestation health, hardware profile, and stake weight.

3. Proof generation

Inference outputs are signed in-enclave and bound to execution context hashes.

4. On chain settlement

Proof references settle on the Layer-1 ledger with fee split and burn routing.

Category multipliers

Reward coefficients prioritize sectors with high societal and institutional value.

Category Multiplier Policy Rationale
Healthcare & Biotech 1.8x Clinical diagnostics and life science acceleration
Scientific Research 1.6x Academic and industrial research workloads
Finance & Risk 1.4x Auditability-sensitive institution workflows
General Inference 1.0x Baseline market computation
Execution Architecture

Hybrid Compute Engine

The engine separates orchestration from execution, utilizing TEEs as the primary high-throughput execution layer alongside optional EZKL zkML co-processing for absolute cryptographic trustlessness.

A

Ingress & policy check

Requests are validated for compliance profile, quota limits, and data residency constraints.

B

Secure dispatch

Workloads route to attested enclaves with matching hardware and cryptographic capabilities.

C

Attested inference

TEE workers run model execution and emit signed output bundles with context hashes.

D

Settlement & Verification

Proofs settle natively via custom EVM precompiles (0x0400 for TEE, 0x0300 for ZKP) for sub-2 second on-chain verification.

Supported confidential compute profiles

Intel SGX, AMD SEV-SNP, AWS Nitro Enclaves, and Azure Confidential VMs housed within sovereign clusters like Stargate UAE, with HSM-backed key custody.

Cryptography

Three layer post quantum security

Cryptographic agility is embedded in the protocol design, not deferred as a future migration burden.

Identity and signatures

ML-DSA-65 (Dilithium 3) secures validator identity, block signatures, and governance attestations.

Transport security

ML-KEM-1024 (Kyber) provides quantum-safe key exchange for validator and client channels.

Integrity anchoring

SHA-256 merkleized state commitments enforce deterministic ledger verification.

Hardware security module support

Validator operations support Thales and Fortanix HSM custody workflows for key isolation, rotation policy, and secure signing ceremonies.

Migration readiness playbook

If cryptographically relevant quantum advances emerge, governance can initiate rapid key policy upgrades, disable deprecated algorithms, and freeze compromised signature families before final state transition execution.

Infrastructure

Validator admission and operating standards

Validator participation requires stake, hardened infrastructure, and policy compliance appropriate for institutional workloads.

Requirement Baseline Notes
Minimum stake 100,000 AETHEL Staking lock establishes economic accountability.
TEE hardware profile SGX / SEV-SNP / Nitro / Azure CVM Remote attestation required for worker enrollment.
GPU class H100 / MI300 equivalent Optimized for 32B parameter inference classes.
Compliance controls KYC/AML + OFAC workflow Mandatory for institutional compute routing.
Uptime target 99.5% monthly Availability impacts reward and slashing profile.

Validator Nodes

Participate in consensus, finality, governance voting, and reward distribution.

Worker Nodes

Execute AI jobs in enclaves and generate attested outputs for settlement.

Sentinel Nodes

Provide DDoS shielding and regional ingress resilience across operators.

North America Europe MENA APAC Sovereign cloud corridors
Need the operator view?

See the dedicated Node Operations guide for role definitions, deployment topology, and production-safe runtime checks.

Benchmarks

Production oriented performance envelope

Measurements reflect sustained testnet workloads with 100+ validator participants and enterprise model mixes.

p50 inference latency 80-150 ms

Optimized 32B-parameter enterprise models in enclave mode.

p95 inference latency 220-340 ms

Enterprise and regulated workload profile.

Model throughput 50,000+ IPS

Aggregate throughput for 32B class models.

State finality <2 sec

Deterministic settlement across validator cluster.

Failover time <12 sec

Region-aware load transfer with worker redundancy.

Network target 2027 1,000+ PFLOPS

Scale path based on validator expansion program.

Trust and Compliance

Security architecture designed for institutions

Aethelred combines cryptographic guarantees with policy enforceability required by regulated sectors.

Protocol controls

Slashing for malicious behavior, timelocked upgrades, attestation revocation, and transparent on-chain governance records.

Attestation verification Circuit breaker hooks Auditability by default