Making AI Sustainable with ReFi

AI Network
4 min readSep 5


Webinar: Sustainable GPU Usage with ReFi
When: Thursday 7th Sept at 5PM EDT.
Sign up:

Join the discussion on how we can make AI more sustainable.

It’s no secret that the world is in a troublesome state when it comes to climate. Carbon emissions have been rising steadily for years as a result of human endeavour — the exploitative effects of industry and consumerism on finite resources have ultimately meant the world is getting hotter, and we all know that comes with its own potentially disastrous set of consequences (not least the potential of human extinction).

The biggest issue is that over decades passed capitalism hadn’t paid much attention to the sustainability of the economies it churned out. Industries have traditionally focused on bottom-line profits, without thinking much about what will happens when the finite resources they use become depleted. This antiquated model is ultimately exploitative, and the majority of industry runners just kicked the can down the road and didn’t think much about what’ll happen when the resources run out in future and there’s nothing left to exploit, and therefore nothing left to profit from.

Carbon emissions are a result of many manmade processes; industry, finance and blockchain included, not least as a result of energy consumption required to power these processes. Of these, AI is no different. The recent surge in AI use and development has led to a significant surge in energy consumption, which is directly causing an increase in carbon emissions.

Artificial Intelligence relies on the power of GPUs (graphics processing units) to function, and as the use of AI inevitably becomes wider, broader and far more encompassing, the amount of GPU processing power needed will rise considerably, as will carbon emissions. Even now with GPT3 (the current LLM that powers the free version of ChatGPT), we can see the enormous consumption of power and the consequent tonnes of CO2 spewed out into the atmosphere.

As Peter Henderson, of The Stanford institute for Human-Centered AI, puts it;

If we’re just scaling without any regard to the environmental impacts, we can get ourselves into a situation where we are doing more harm than good with machine learning models. We really want to mitigate that as much as possible and bring net social good.

This is the essence of Regenerative Economics, a term coined by John Fullerton in 2015, which in its most basic form is the notion of responsibly utilising resources without exhausting them, or harming other parts of the broader societal or environmental system, and allowing/helping them to regenerate so as not to bankrupt the ecosystem on which an economic process relies.

A simple example is that of industrial scale fishing. If fisherman overfish and clear out a species in a certain area before the species has had the chance to reproduce enough to replenish their numbers, the fisherman exhaust the supply of their product and bankrupt their economic system. This often occurs in the pursuit of short term profits, without much attention to long term sustainability, which leads not only to the bankrupting of the exploited resource, but also of the exploiter themselves.

This is what we wish to avoid with AI — a state where the use of AI contributes so much to carbon emissions that it significantly exhausts the health of the global environment — and we can do this with Regenerative Economics in its Web3 application; Regenerative Finance (ReFi).

ReFi seeks to utilise the blockchain to decouple the creation of monetary value from the traditionally unsustainable methods of finance, and aims to solve economic, social or environmental issues uses regenerative blockchain-based cycles.

This is exactly what we here at AI Network are aiming to achieve; to contribute to offsetting carbon emissions of Artificial Intelligence using ReFi.

We’re an AI development ecosystem, built on its own layer 1 blockchain, which gathers GPU resources for the creation and development of AI applications, and we can contribute to the offsetting or carbon emissions in a few ways.

Firstly we can connect our GPU resources to other regenerative projects, directly contributing processing power to projects focused on sustainability. To this end we can also collect idle GPUs from resource providers to power AI ReFi projects, eliminating waste of idle processing power not being used anywhere else. We can incentivise those with idle GPUs to contribute their processing power to ReFi projects with increased rewards in $AIN token (AI Network’s native crypto token).

Secondly we can acquire carbon-free GPUs; processing chips powered by carbon free energy sources that minimise environmental impact, which we can then put toward carbon neutral and sustainability projects, further reducing emissions.

Thirdly we can provide GPUs or funds sourced from GPU NFT sales to carbon offsetting projects, resulting in environmental assets like carbon credits. If these credits are then sold for profit, more token rewards can be shared with GPU providers contributing their resources to ReFi projects, hence incentivising providers further to contribute to sustainability initiatives.

At AI Network we believe in green AI, and we will do our utmost to ensure AI use and development becomes as sustainable as possible. To this end, we are planning to transition all GPU resources within our ecosystem to carbon-free resources by the end of the year.

Work with us for a greener future in AI. Become a partner and utilize our GPU processing capabilities to help power your sustainability projects.

Tune in to the discussion about sustainable AI in our webinar on Thursday 7th Sept at 5PM EDT.

Sign up here:



AI Network

A decentralized AI development ecosystem built on its own blockchain, AI Network seeks to become the “Internet for AI” in the Web3 era.