decentralised Compute to push the boundaries of machine learning

The Gensyn network is the Machine Learning Compute Protocol that unites all of the world’s compute into a global supercluster, accessible by anyone at any time

Sign up to our newsletter

Backed by

Arise Health logoArise Health logo2020INC logoArise Health logoThe Paak logoArise Health logoThe Paak logoOE logo

A Solution for Scale

Deep learning models are trained across devices that have joined the network

This is some text inside of a div block.
This is some text inside of a div block.
This is some text inside of a div block.
This is some text inside of a div block.

Easy to Provision

Headline goes here, headline goes here, headline goes here.

Copy lines here. Copy lines here. Copy lines here. Copy lines here.

Hyperscale

On-demand supercompute

Devices submit proofs to the Gensyn blockchain to prove that the model has been trained as requested.
All devices in the world can join the network.

UNCENSORABLE

“We cannot have a new clergy who get to interpret the gospel and decide whether or not a layperson gets access”

Edward Snowden speaking to Gensyn about why AI training should be decentralised and permissionless (2022)

FICIENT************COST*EFFICIENT********COST*EFFICIENT********COST*EFFICIENT**

Peer-to-Peer; no margins

The Gensyn protocol disintermediates cloud oligopolists like AWS and does not charge a margin. This sharply lowers the cost of compute. More here.

****HYPERSCALE************HYPERSCALE*********HYPERSCALE*********HYPERSCALE**

Universal supply of devices

As a P2P network, any device in the world can connect to Gensyn. This includes underutilised devices like consumer GPUs, custom ASICs, and SoC devices which are capable of training neural networks.

UNCENSORAPERMISSIONLESS************PERMISSIONLESS*********PERMISSIONLESS*********PERMISSIONLESS**

“We cannot have a new clergy who get to interpret the gospel and decide whether or not a layperson gets access”

Edward Snowden speaking to Gensyn about why AI training should be decentralised and permissionless (2022).

Featured in