What you need to know 🐳 Io.net fits into the decentralised physical infrastructure (DePIN) narrative and builds on existing networks.
The IO token is based on the Solana blockchain, which is attracting many DePIN projects thanks to its low costs and high transaction throughput.
Decentralised cloud computing is cheaper than centralised cloud computing, but it does not seem suitable for certain tasks, including training AI based on Large Language Models (LLMs).
Overview 🧬 Io.net is an aggregator of IT components that offers decentralised cloud computing services.
Anyone, private or professional, can join the IO network to rent or lease computing power from IT hardware. This computing power can be used for a variety of tasks, including machine learning.
IO sees computing power as the digital "oil" fuelling today's technological revolution and is looking to best connect supply and demand for this new resource.
IO's aim is to compete with centralised cloud providers such as Amazon Web Services or Microsoft Azure, by offering more attractive prices and greater flexibility in its offerings.
IO is part of the DePIN (decentralised physical infrastructure) project sector and builds on other projects in the same sector. In addition to its own network of graphics cards and processors, IO also incorporates Rendr's network and Filecoin's storage capabilities.
DePIN projects are becoming increasingly popular because they offer services to the whole world and are not in closed cycles like most crypto projects.
The major difficulty with DePIN projects is successfully connecting supply and demand. The strategy used by almost all of these projects is first to set up a consistent supply through incentives paid for in their own token.
Financing 💰 Io.net raised a total of $30 million in a round announced in March 2024 from Hack VC, Multicoin Capital, Delphi Ventures, Animoca Brands, OKX Ventures and Solana Ventures.
Team and community 👾 The project team is primarily based in New York.
Ahmad Shadid, the founder and former CEO of IO, announced his departure on 9 June. He is now developing Obot, a network of autonomous agents based on IO.
Tory Green, former COO, is now IO's new CEO. He has been a Partner at venture capital fund Tiller Partners since 2018.
Gaurav Sharma is IO's CTO. He previously worked as an engineer at Binance and Agoda.
The project has 540,000 followers on twitter and 550,00 members on Discord.
Functioning ⚙️ The IO ecosystem consists of the following:
io.net: the company that maintains the IOG network and builds products and services based on it. IOG (Internet of GPU) network: a decentralised physical infrastructure made up of geographically distributed computing hardware, providing authorisation-free access to computing resources on demand. IO Cloud: the platform enabling users to virtualise scalable and configurable clusters of the IOG network on demand. IO Cloud offers great variety and flexibility, with the ability to choose the models of the IT components to be rented, their location, the latency between them, etc.
Limits 😨 Large Language Models (LLM)-based AI training seems to have significant limitations on decentralised networks , as this type of task requires very large quantities of graphics cards working in parallel with almost non-existent latency.
The AIs of the largest companies in the sector are trained in specialised data centres with a very sophisticated infrastructure in terms of connectivity, cooling, etc. These conditions cannot be replicated within a decentralised network.
This is not to say that decentralised cloud computing networks are useless in the AI sector, since they can in particular be used to perform inference, i.e. use AI models that have already been trained.
The network has already suffered an attack in the past.
The IO token 🪙 IO is the token at the heart of the project and is based on the Solana blockchain.
There is a maximum quantity set at 800 million, with currently 95 million in circulation.
Payments to use Io.net services can be made in fiat, USDC, or any other token supported by the network to facilitate access for all users. However, they are then automatically converted into IO with a 2% fee that burns IO tokens. It is possible to pay directly in IO to avoid these fees.
A total of 300 million IO will be distributed to computing power providers and IO stakers every hour for 20 years.
The rewards follow a disinflationary model, starting at 8% in the first year and decreasing by 1.02% per month (~12% per year) until they are all distributed.
The IO token is distributed as follows:
50% for the community including 5% dedicated to the launch airdrop 16% for Io.net's research and development 11.3% for initial contributors 22.7% for private investors Regulations ⚖️ Like most projects that raise funds in tokens, Io.net is exposed to the risk of being reclassified as an unregistered financial security. On the other hand, this does not pose a problem under European regulations.
Roadmap 📝 IO plans to improve its offering with the integration of data anonymity features and improvements to the way clusters work. New models will be introduced to enable new use cases such as cloud gaming.
Competitors ⚔️ Akash Network is another decentralised cloud computing project that is older than IO, but does not aggregate other networks.
Gensyn is a future aggregator of DePIN decentralised cloud computing projects that has raised a total of $50.6 million from a16z among others. However, unlike IO, Gensyn wants to focus specifically on machine learning.
Where can you buy the IO token? 🛒 It is possible to buy the IO token on Binance or on the DEX Meteora on Solana.
The Big Whale's opinion 🐳 IO is part of a very competitive decentralised cloud computing market.
It has the advantage of positioning itself as an aggregator of other projects of the same type thanks to its partnerships.
its strength lies in focusing on the distribution and optimisation of the cloud computing offering.
The increase in general demand for computing power over the years is almost certain, however it is not certain that decentralised cloud computing can compete with centralised data centres on all tasks.
