What is io.net (IO)?
👋 New to this? Just start reading at the top — it begins in plain English and gets more detailed as you scroll. Jump to any level:
🟢 The simple version
Plain English — no jargon. Start here.
io.net is a decentralised GPU network where anyone with a spare GPU — from a gaming PC to a data centre server — can rent it out to AI companies and developers who need computing power to train machine learning models, with payments handled by the IO token.
Why does the world need a decentralised GPU network?
Training an AI model requires enormous computing power — specifically GPU (Graphics Processing Unit) compute. The problem: most of that GPU capacity is locked up with AWS, Google Cloud, and Microsoft Azure, who charge high prices and often have waitlists. Meanwhile, there are millions of GPUs sitting idle — in gaming PCs, crypto mining farms (whose RTX cards are now less profitable for mining), university labs, and private data centres. io.net connects these idle GPUs to AI developers who need compute, cutting out the hyperscalers.
This is the DePIN (Decentralised Physical Infrastructure Network) model applied to compute: instead of a company owning all the hardware, a token incentivises individuals and businesses to contribute hardware to a shared network. The contributor earns IO tokens. The AI developer gets cheaper, faster access to GPU clusters.
IO token
IO is the native token of io.net. GPU suppliers are paid in IO for contributing compute. AI developers pay IO to access GPU clusters. IO also serves as a governance token for the network. The token launched in June 2024, coinciding with io.net's mainnet launch and airdrop to GPU contributors who had participated in the testnet phase.
Is IO legal in India?
Yes. IO qualifies as a Virtual Digital Asset (VDA) under Indian law. A flat 30% tax applies to gains, and 1% TDS applies to qualifying trades on Indian exchanges. If you earn IO by contributing GPU compute, those earnings are also taxable income in India — seek guidance from a tax professional on reporting compute income.
🟡 A bit more detail
For when you want to go a little deeper.
How to supply GPU compute on io.net
To earn IO by contributing compute: you run the io.net worker software on a machine with a compatible GPU (NVIDIA RTX 3080 or better, or data-centre grade A100/H100s). The software connects your GPU to the io.net network. When a developer requests a GPU cluster, io.net allocates your GPU, you earn IO tokens for the uptime and compute delivered. Earnings depend on your GPU grade, uptime, and current network demand. High-end GPUs (A100, H100) earn significantly more than consumer cards.
io.net vs AWS/Google Cloud
io.net positions as 90%+ cheaper than hyperscaler GPU pricing. AWS H100 instances can run $30–$50/hour per GPU; io.net targets $1–5/hour ranges. The tradeoff: reliability and consistency guarantees are lower than managed cloud services. For burst AI training workloads where cost matters and the job can tolerate some variability, io.net is compelling. For production inference serving that needs 99.9% uptime SLAs, hyperscalers still win.
io.net's value depends on sustained AI training demand and whether AI developers actually use decentralised compute at scale. Many large AI labs have long-term cloud contracts and prefer reliability over cost savings. Real utilisation of the network (not just connected supply) is the key metric to watch. Live data: CoinGecko
🟣 The full technical picture
For the technically curious.
Key facts
- Token: IO (payment + governance)
- Total supply: 800 million IO
- Mainnet / token launch: June 2024
- Category: DePIN — Decentralised Physical Infrastructure Network
- What contributors supply: GPU compute (NVIDIA GPUs primarily)
- What buyers access: GPU clusters for AI/ML training workloads
- Claim: 90%+ cheaper than AWS/GCP equivalent GPU compute
- Compatible hardware: Consumer GPUs (RTX 3080+) to data-centre (A100, H100)
- Chain: Solana (for payments and coordination)
- Backing: Multicoin Capital, Hack VC, Solana Ventures; raised ~$40M
- Competitors: Render Network (RNDR/RENDER), Akash Network (AKT)
DePIN compute — how the cluster allocation works
When an AI developer requests a GPU cluster on io.net, the network's matching engine allocates GPUs from the contributor pool based on: GPU type, geographic location (for latency), uptime history, and availability. Multiple GPUs can be combined into a virtual cluster. Payments are streamed continuously as compute is delivered — not post-hoc. The IO token is the settlement layer. Network proofs verify that compute was actually delivered (not just claimed), using cryptographic verification of task completion.