Skip to content

Beam-Network/beam

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BEAM

An open coordination layer for bandwidth and machine-to-machine data transfer.

Why BEAM Exists

The modern internet has developed large-scale markets for compute and storage, but not for bandwidth.

Cloud platforms provide on-demand compute. Storage networks allow data to be stored and retrieved globally. However, the movement of data — bandwidth — remains controlled by centralized providers: cloud platforms, CDNs, and telecom operators.

As a result:

  • Bandwidth pricing is opaque and expensive
  • Unused global network capacity cannot participate in data delivery
  • Routing decisions are controlled by centralized infrastructure

Despite the enormous scale of global networking infrastructure, there is no open coordination layer for bandwidth.

BEAM is designed to address this gap.

The Missing Internet Layer

Compute has markets. Storage has markets. Bandwidth markets exist, but they are largely closed, centralized, and inaccessible to independent participants.

BEAM introduces an open coordination layer where bandwidth can be contributed, measured, and rewarded based on performance.

What BEAM Does

BEAM aggregates distributed bandwidth and coordinates data transfers across a network of participants.

  • Nodes contribute bandwidth and participate in executing transfers
  • Proof-of-Bandwidth measures and validates bandwidth performance during transfers, allowing validators to assess delivery quality
  • Rewards are based on measurable performance, not promises

Transfer Patterns

BEAM supports flexible source-to-destination configurations:

Pattern Description
Single → Single One source to one destination
Single → Multi One source replicated to multiple destinations (fan-out)
Multi → Single Multiple sources aggregated to one destination
Multi → Multi Multiple sources to multiple destinations (mesh)

Each transfer is chunked and distributed across workers in parallel.

Capabilities

  • Universal Storage Connectivity — S3, GCS, Azure Blob, R2, IPFS, HTTP endpoints
  • Multi-Destination Fan-Out — Replicate data to multiple destinations in a single transfer
  • Parallel Transfers — Large files split into chunks and transferred in parallel
  • Webhook Callbacks — Get notified when transfers complete
  • MCP Integration — AI agents can invoke BEAM as a tool for data movement

Built for the Machine Internet

Machine-to-machine communication is rapidly increasing:

  • AI-to-AI — Autonomous agents exchanging data with AI services
  • Agentic workflows — AI systems coordinating tasks and sharing outputs
  • AI training pipelines — Datasets moving to compute, models distributing to inference
  • Data center synchronization — Replication across clusters and regions
  • AI-to-IoT — Sensor data flowing to ML systems

These workflows involve terabytes or petabytes moving across regions and independent systems. BEAM provides a decentralized coordination layer for this data movement.

How It Works

┌──────────┐                                                    ┌─────────────┐
│  Client  │ ── transfer ──▶  ┌──────────────┐  ── tasks ──▶    │   Workers   │
└──────────┘                  │   BEAMCORE   │                  └──────┬──────┘
                              │ (Coordinator)│                         │
┌──────────────┐              └──────┬───────┘                         │ delivery
│ Orchestrators│ ◀── assignments ────┘                                 ▼
│   (Miners)   │ ── proofs ──────────▶                          ┌─────────────┐
└──────────────┘                                                │ Destinations│
                              ┌──────────────┐                  └─────────────┘
                              │  Validators  │ ◀── read proofs ──┘
                              └──────────────┘
  1. Clients create transfers via BeamCore API or SDK
  2. BeamCore assigns chunks to orchestrators by stake-weighted allocation
  3. Orchestrators poll BeamCore for assignments and create tasks
  4. BeamCore pushes tasks to workers via WebSocket
  5. Workers fetch data from source and deliver to destinations
  6. Workers submit Proof-of-Bandwidth to BeamCore
  7. Validators read proofs from BeamCore and set weights on chain

Note: BeamCore is the central coordination layer. Data flows directly between sources, workers, and destinations.

The Vision

BEAM's goal is to become the open bandwidth layer for the machine internet.

By coordinating distributed bandwidth and measuring performance through Proof-of-Bandwidth, BEAM enables data to move efficiently across a network that is:

  • Open to participation — Contributors join via the Bittensor network
  • Performance-based — Rewards tied to measured delivery metrics
  • Decentralized — No single point of control

Run a Node

This repository contains the code for running orchestrators and validators on the BEAM network.

Network Information

Network Subnet UID Subtensor BeamCore URL
Testnet 304 test https://beamcore-dev.b1m.ai
Mainnet 105 finney https://beamcore.b1m.ai

Links

License

MIT License — see LICENSE

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors