Q4 202501/05
AETRON
Launch your AI service on a decentralized network — its participants provide the compute. Or connect your GPUs to the network's workloads and get rewarded for the work.
The AETRON LLM Pre-Alpha is Live on the Mainnet! The LLM has launched on the AETRON decentralized network.
What AETRON is
Thousands of GPUs around the world are joined into a single network. Users launch AI services on it, and GPU owners contribute their compute to run them. The network checks every computation itself and pays out for the work.
How this differs from ordinary mining
POW
Ordinary mining (Proof-of-Work)
Graphics cards spend all day guessing random numbers, only to prove that electricity was spent on it. Nobody ever needs the result. Around 170 TWh a year — more than Norway consumes.
POI
Mining on AETRON (Proof of Intelligence)
Graphics cards do real work: answering requests to AI models, training them, running AI agents. Every completed job leaves a proof behind, and any participant in the network can re-check it.
Useful work without new consumption
The network uses GPUs that are already running: at home, in offices and in datacenters. Their owners are paid for the work delivered, not for the kilowatts burned.
How the network proves that a computation was done honestly — the mechanics are on a page of their own.
What AETRON Brings
A distributed network worth participating in for everyone
01
For Network Participants
- Earn by providing computational power
- Receive AETRON tokens for solving AI tasks
- Become part of global AI infrastructure
02
For Developers and Business
- Access powerful AI computing without million-dollar investments
- No dependency on pricing and rules of large corporations
- Launch your own neuronets for your own AI tasks
03
For the Whole World
- AI technologies become open and accessible
- Compute capacity goes into productive work
- No one can control or shut down the network
You can join in one of several roles: neuronet owner, miner, validator or network node.
An infrastructure layer for AI
AETRON shifts AI development toward distributed infrastructure. Miner capacity goes into AI work, without the carbon trail of Bitcoin.
01
Open AI access
AI capability is concentrating in 5–7 of the world's largest corporations. AETRON opens access to frontier AI through a distributed compute base - without dependency on a single supplier.
02
Accelerated development
Distributed training and inference remove the capital barrier. AI development speeds up through the joint work of thousands of network participants.
03
Proof of Intelligence
Proof of Intelligence consensus delivers real AI capability with transparent verification on every unit of work.
04
A new AI economy
AI capability creates value for the participants of the network. New standards take hold - for instance, AIfi.
Be part of a new standard for AI
A network for distributed AI
AETRON's architecture is built to keep the network stable across thousands of concurrent participants and a distributed compute base.
poi
Proof of Intelligence
The consensus mechanism verifies real AI capability through cryptographic proofs.
Learn MoreneuronetsNeuronets
A neuronet is a standalone AI service on the AETRON network with three roles: owner, miners, stakers. Emission splits automatically by confirmed work.
Learn MoreminingAI Mining
A single AETRON client binary. Supports Apple Silicon, Ampere, Ada, Hopper, Blackwell and CDNA3 architectures. One command to start.
Learn MorechainBlockchain Infrastructure
12-second blocks, Merkle tree for batched proofs, governance currently on Proof of Authority, transitioning to voting by validators.
Learn MoreconstructorConstructor
A console for launching Neuronets without writing a line of code. You pick the model, the tasks and how you want to earn, and everything that touches the chain is handled for you.
Learn MoreagentsAI Agents
Autonomous AI agents that solve multi-step tasks and work with tools. Access via the internal orchestrator or the OpenAI-compatible endpoint.
Learn Moreaetron · preview activity
[ 09:41:02 ] peer connected 4 812 online
[ 09:41:03 ] task #98 121 → miner 5Grw…utQY H100
[ 09:41:04 ] inference done · 1.9s logits ✓
[ 09:41:04 ] peer-replay → verdict: honest
[ 09:41:05 ] proof batched → merkle 8f3a… 412 tx
[ 09:41:06 ] block #182 449 sealed 12.0s
[ 09:41:06 ] emission → miners · stakers AET
[ 09:41:07 ] peer connected 4 815 online
[ 09:41:08 ] task #98 122 → miner 5FHn…94ty MI300X
[ 09:41:09 ] training step 40 128 · loss ↓ ckpt ✓
[ 09:41:10 ] peer-replay → verdict: honest
[ 09:41:11 ] diffusion · 28 steps · latents commit ✓
[ 09:41:12 ] proof batched → merkle c4e1… 398 tx
[ 09:41:12 ] block #182 450 sealed 12.0s
[ 09:41:13 ] peer connected 4 821 online
[ 09:41:14 ] task #98 123 → miner 5HTE…8Mxq RTX 5090
Live since October 2025
Network status
PoI deployment
Current phase
12 seconds
Block time
6 GPU architectures
Hardware support
[Roadmap]
Building the future
Why the network is built this way — in the project manifesto.
