io.net is a decentralized computing platform that enables users worldwide to access a network of GPUs, providing scalable computational resources on demand. The platform operates across 138 countries and is designed to make high-performance computing for applications like artificial intelligence (AI) and machine learning (ML) accessible through a distributed cloud model. ^[1][2]

Overview

io.net functions as an alternative to traditional centralized cloud providers. Through its distributed infrastructure, IO Cloud, users can deploy and manage GPU clusters with reduced need for dedicated hardware or complex management tasks. Built on the Ray framework, IO Cloud integrates efficiently with machine learning and AI projects, offering flexibility, scalability, and resilience. ^[1][2][3]

Key Features

1. On-Demand GPU Access
   io.net’s on-demand model provides access to GPU resources without permission barriers, with tools to streamline deployment and enable developers to pre-configure and manage their ML stacks efficiently.
2. Global Infrastructure
   Operating in over 138 countries, io.net supports a broad network of GPUs and CPUs, with distributed infrastructure to reduce latency and maintain performance for AI and ML applications.
3. Hardware Options
   io.net offers a selection of AI processors and flexible processor configurations, with scalable clusters that automatically adjust to fluctuating user demand.
4. Sustainable and Secure Operations
   Through partnerships with certified data centers, io.net provides sustainable GPU access options. The platform is SOC2-compliant and uses end-to-end encryption to secure data.
5. Integrated Payment System
   io.net is integrated with the Solana blockchain, enabling payments through Solana Pay with minimized fees for USDC transactions (2%) and a 0.25% reservation fee for compute resources.

IO Workers

The IO Workers program allows users to lend their GPUs and CPUs via a web application, contributing to a decentralized network of computational power. IO Workers is equipped with monitoring tools to oversee real-time performance and resource use. ^[4]

Key Features

• Scalability and Flexibility: Allows for adjustable compute resources based on workload demands.
• Security and Control: Ensures secure resource sharing with controlled access.
• Global Access: Provides worldwide accessibility for various industries and regional requirements.

IO Explorer

The IO Explorer is a monitoring tool designed for insight into io.net’s network metrics and operational data. This feature provides detailed views of cluster performance, resource utilization, booking rates, and transaction history, allowing for an evaluation of network functionality in real time. With a focus on data privacy, IO Explorer restricts access to non-essential information, displaying only relevant system metrics to authorized users. ^[5]

IO ID

IO ID acts as a centralized hub within the io.net ecosystem for managing finances associated with resource contributions. It organizes information on earnings and expenses, facilitating structured oversight of revenue from GPU and CPU contributions. The tool provides a clear summary of financial transactions and supports cryptocurrency withdrawals, for managing the income generated within the network. ^[6]

IOG Network

The IOG Network, known as the "Internet of GPUs™," extends io.net’s decentralized computing capabilities. Leveraging a customized Ray framework, it optimizes workload distribution and enhances scalability to support high-performance applications, including confidential computing and large-scale AI training. As an open-source platform, the IOG Network promotes collaborative development within the global AI community. ^[1][2]

IO Token

The IO token is integral to the io.net ecosystem, operating as a unit of exchange and a tool for network incentives. Designed to support a distributed infrastructure for artificial intelligence (AI) and machine learning (ML) applications, IO enables interactions among participants, particularly those providing and consuming GPU and CPU resources. As demand for scalable computing resources grows, IO provides a framework to distribute resources efficiently while encouraging continued participation from network contributors. ^[7]

Tokenomics

Initial Supply

At launch, io.net released 500 million IO tokens as the initial circulating supply, with a total cap of 800 million tokens. Distribution is scheduled over 20 years, following a structured release model that aims to provide a balance between availability and gradual supply reduction. The token supply limit is set at 800 million, with a planned reduction in emission rate over time as the network evolves.

Disinflationary Emission Model

Following a disinflationary schedule, similar to models used by other networks, io.net will gradually reduce IO’s annual inflation rate. Starting at an initial rate of 8%, this inflation rate will decrease approximately 12.165% each year, reaching about 0.68% by the end of the 20-year period. This model aims to release tokens sustainably, gradually lowering inflation as network demands stabilize.

Hourly Emissions Across Epochs

IO emissions are structured as hourly releases over 175,319 epochs, or hourly periods, across 20 years. This predictable release model enables a steady token supply reduction over time. For example, the first epoch releases 4,566.21 tokens in the initial hour. By the next hour, emissions decrease to 4,566.15 tokens, with reductions continuing at a gradual rate as the inflation rate lowers incrementally.

Emission Illustration for Initial Epochs

The initial epochs illustrate how IO emissions are designed to decrease gradually:

• Epoch 1: 4,566.21 tokens, with an inflation rate of 0.00091324%
• Epoch 2: 4,566.15 tokens, with an inflation rate of 0.00091323%
• Epoch 3: 4,566.08 tokens, with an inflation rate of 0.00091322%

This gradual reduction in emissions is structured to align with the network's growth and resource demands over time. ^[2][7]

Staking

Staking is a fundamental part of io.net's tokenomics, developed to help ensure GPU and CPU suppliers maintain network standards. Participants are required to stake IO tokens according to the hardware capacity they bring to the network, with a base stake of 200 IO per GPU, subject to adjustments based on the device’s earning multiplier.

• Staking Requirements: Suppliers must stake a specified amount of IO per device, determined by the device’s capacity and earning potential.
• Rewards Distribution: Block rewards are distributed based on each supplier’s staked amount, with funds held securely in smart contracts.
• Unstaking Process: A 14-day cooldown is triggered when a supplier chooses to unstake, during which staked tokens are ineligible for block rewards. Unstaking actions cannot be reversed until the cooldown period concludes.

To maintain network integrity, io.net applies a slashing mechanism for any supplier involved in actions deemed malicious or harmful, such as service inconsistency or falsified data. If a violation is confirmed, a portion of the staked IO is slashed, following a reconsideration period of one month. During this period, suppliers may appeal against the penalty. If the appeal is unsuccessful, the slashed tokens are removed from circulation.^[1][8]

Future Expansion of Staking Options

Currently, io.net’s staking program is restricted to device owners, who may stake IO directly to their own hardware. Future development phases may introduce options for non-node operators to stake, expanding access to rewards and governance participation across the network.

Leadership Transition

Shortly before io.net’s token launch on June 11, 2024, co-founder and CEO Ahmad Shadid stepped down from his role, effective immediately. Following his departure, co-founder and former Chief Operating Officer Tory Green assumed the role of CEO. Io.net, a decentralized infrastructure provider on the Solana blockchain, enables GPU rentals for machine learning and AI workloads, creating a network intended to provide cost-effective computing power.

Shadid’s departure coincided with community concerns about previous statements regarding io.net’s GPU resource availability. Although he did not directly address these concerns, Shadid stated that his decision was meant to allow io.net to advance without potential distractions. This transition also followed a GPU metadata attack in April that temporarily reduced active GPU connections from 600,000 to 10,000.

In response to community speculation about his token holdings, Shadid clarified that his tokens were under a four-year lockup, with team and investor tokens restricted from sale until June 2025. Additionally, he pledged 1 million IO tokens to the Internet of GPUs Foundation to support ecosystem growth.

New CEO Tory Green described the token launch as a critical milestone in io.net’s development toward building a decentralized AI compute network. Green noted that io.net operated approximately 20,000 cluster-ready GPUs for AI inference and model training workloads, with ongoing efforts to expand supplier and user participation. ^[9][10]