io.net (IO)

Overview

What is io.net?

io.net (ticker: IO) is a decentralized computing network built to connect underused graphics cards (GPUs) around the world and make them available to people who need computing power for artificial intelligence, machine learning, and other heavy workloads. Instead of renting from a single cloud company, users can spin up clusters of GPUs from many independent providers and run their jobs across them. The project is built on the Solana blockchain and grew out of the Solana developer community. Its goal is to make “compute” a resource you can access like electricity—on demand, at scale, and at lower cost than traditional cloud platforms. (io.net)

What problems does it try to solve?

Modern AI models need large amounts of parallel compute. Centralized clouds are often expensive, booked out, or limited to certain regions and GPU models. io.net aggregates supply from data centers, crypto miners, and other hardware networks (such as Filecoin or Render) and combines them into a marketplace. Developers can then reserve clusters, run training or inference, and pay through a simple portal, while suppliers earn for sharing their idle GPUs. (io.net)

History & Team

Origins and milestones

The idea behind io.net took shape during Solana hackathons and builder events in 2023. The team pushed toward a public beta later that year and began framing the network as an “Internet of GPUs,” a decentralized layer that coordinates compute supply and demand. Over time, the documentation and product suite expanded to include a portal for renters, an interface for GPU suppliers (Workers), and an explorer for network data. (io.net)

Founders and leadership

The founding group includes Ahmad Shadid, Tory Green, Basem Oubah, and Matej Tomazin. In June 2024, ahead of IO’s token launch, Shadid stepped down as CEO, and co‑founder Tory Green took over leadership. Public reporting at the time noted the immediate transition and the handoff to Green as the project prepared for its next phase. (theblock.co)

Backers and funding

In March 2024, IO Research—the company building io.net—announced a $30 million Series A round led by Hack VC, with participation from Multicoin Capital, 6th Man Ventures, Solana Ventures, OKX Ventures, Aptos Labs, Delphi Digital, and others. This round helped fund expansion of the decentralized GPU cloud and growth of the supplier network. (theblock.co)

Technology & How It Works

Multi‑layer architecture

io.net’s stack looks like a modern, cloud‑style system designed for distributed AI work. The portal sits on a multi‑layer architecture with a security layer (firewalls, authentication, logging), an API layer (FastAPI, GraphQL, REST), a backend layer (Python/Node services plus an IO‑SDK), and an infrastructure layer that handles orchestration, execution, storage, and monitoring. Tools include Kubernetes and Airflow for orchestration; Ray, PyTorch, TensorFlow for execution; Docker for containers; and S3/HDFS for storage. This design lets the network spin up and tear down GPU clusters for jobs quickly. (io.net)

IO‑SDK and distributed computing

A key software piece is the IO‑SDK, a customized fork of Ray (a popular open‑source system for distributed Python). Ray allows tasks to run in parallel and share data in memory. io.net’s fork focuses on auto‑scaling, fast task scheduling, and integration with common ML frameworks, making it easier to spread training, inference, or reinforcement learning jobs across many GPUs. (io.net)

Networking: mesh VPN and tunnels

To link many machines across geographies, the network uses a secured mesh VPN. In a mesh, nodes connect directly to each other instead of through a single hub. This helps reduce latency, improve resilience, and spread traffic loads. io.net also uses reverse tunnels to reach Workers behind firewalls or NATs, making it simpler for engineers to manage devices securely without complex network setup. (io.net)

Job orchestration and workflows

Through the portal, teams can reserve GPU clusters and run common AI workflows:

• Batch inference and model serving for deployed models
• Parallel model training and hyperparameter searches
• Reinforcement learning jobs across distributed workers The service is designed for Python workloads and aims to minimize changes needed to move from a single machine to a large distributed cluster. (io.net)

Tokenomics & Utility

IO as the network’s economic engine

IO is the native token of the IOG (Internet of GPUs) Network. It is used to coordinate supply and demand, compensate suppliers, and align incentives around reliable compute. Users may pay for compute in USDC or IO, but suppliers are compensated in IO, which creates structural token demand with each job. Paying in 100% IO also removes certain payment fees compared with paying in USDC. (io.net)

Supply, emissions, and rewards

• Fixed maximum supply: 800 million IO.
• Genesis supply: 500 million at launch.
• Emissions: 300 million emitted as hourly rewards to suppliers and their stakers over 20 years, using a disinflationary schedule that starts near 8% annual inflation in year one and decreases each month until the cap is reached. Rewards are credited hourly and may be subject to slashing before distribution. Some supplier participation requires a minimum stake to qualify for block rewards. (io.net)

Fees and programmatic burns

io.net charges small reservation and payment fees to both renters and suppliers. Payments in USDC incur a 2% payment fee, while payments in 100% IO have no payment fee. The network uses revenue to buy and burn IO on a programmatic basis, introducing a deflationary element tied to actual usage. (io.net)

The Incentive Dynamic Engine (IDE)

In addition to the baseline model above, io.net introduced the Incentive Dynamic Engine, which aims to tie token flows to real demand. The IDE targets more predictable, USD‑denominated supplier payouts to offset token volatility; after supplier payments, at least half of remaining revenue is burned, with a stated long‑term goal to remove a large portion of the supply over time. The system adjusts payouts based on metrics like utilization so incentives follow actual network activity. (io.net)

Ecosystem & Use Cases

Who uses io.net?

• AI startups and ML teams that need clusters of GPUs for training and fine‑tuning models
• Developers running high‑throughput inference and model serving
• Builders of reinforcement learning systems that scale across many workers
• Projects experimenting with consumer apps that need GPU backends (for example, BC8.ai used io.net to generate and serve AI images) (io.net)

How suppliers participate

Independent data centers, mining operations, and professionals with compatible GPUs can connect devices as Workers. They earn IO from jobs and from block rewards that encourage continuous, reliable supply. Dashboards and APIs show earnings, jobs served, and cluster activity. (io.net)

Tools for builders

The ecosystem includes:

• A portal for reserving compute and managing spend
• An IO Explorer with APIs for network data
• Guides for deploying distributed training and inference on the platform These resources help teams move from small tests to large clusters without reinventing their stack. (io.net)

Advantages & Challenges

Advantages

• Access to a global pool of GPUs: Aggregation lets users find capacity when centralized clouds are tight.
• Flexible orchestration: With Kubernetes, Ray, and familiar ML frameworks, jobs can scale horizontally.
• Demand‑linked incentives: The IDE and fee structure push rewards to follow real usage, not just emissions.
• Payment design: Paying in IO removes certain payment fees and every job creates structural IO demand via supplier compensation. (io.net)

Challenges

• Heterogeneous hardware and networks: A decentralized pool can vary in latency, bandwidth, and GPU models; scheduling and verification must handle this variety.
• Competition with hyperscalers and other DePIN projects: Centralized clouds still offer deep integrations and global footprints; DePIN networks must win on cost, availability, and openness.
• Operational complexity: Coordinating mesh networking, reverse tunnels, job retries, and data movement across many providers is non‑trivial and must be continuously improved. (io.net)

Where to Buy & Wallets

Exchanges and DEXs

• io.net can be purchased on Binance. The token went through Binance Launchpool and then opened for trading with multiple pairs. (cointelegraph.com)
• IO is available on KuCoin. (kucoin.com)
• IO is available on Bybit (IO/USDT). (announcements.bybit.com)
• IO is available on Gate.io (including pre‑market support ahead of spot listing). (gate.com)
• IO is available on Solana DEXs such as Raydium through routers like Jupiter; an IO/USDC pool exists on Raydium’s CLMM. (geckoterminal.com)

Note: Availability can vary by region and account type. Always check the exchange’s official listing page for supported pairs in your jurisdiction.

Wallet support

• IO follows the Solana SPL token standard. You can store IO in popular Solana wallets such as Phantom, Solflare, and Backpack, and with hardware devices like Ledger when used via Solana‑compatible wallet apps. Solflare and Phantom publicly document support for SOL and SPL tokens, and Backpack provides self‑custodial Solana support. (solflare.com)

Regulatory & Compliance

Data and platform standards

io.net states alignment with common enterprise security and privacy frameworks in its public materials, including references to SOC 2 and GDPR on its tokenomics page. These references speak to process and controls rather than blockchain regulation, but they show an intention to meet recognized standards for handling user data and operations. (io.net)

Jurisdictional considerations

The IO token functions as a utility token within a compute marketplace: it coordinates payments, rewards, and staking within the network. As with many crypto projects, classification can vary across countries and over time. Centralized exchanges that list IO apply their own listing reviews and KYC/AML rules for customers in supported locations. Because io.net runs on Solana and settles value on‑chain, users typically interact through compliant exchanges or self‑custodial wallets depending on local rules and the nature of their activity. (cointelegraph.com)

Shariah/halal treatment

io.net is not considered shariah compliant by any widely recognized Shariah Supervisory Board at this time. Publicly available materials do not show a formal certification or fatwa for IO. From an Islamic finance perspective, some scholars look for clear asset‑backing, avoidance of interest (riba), gambling (maysir), and excessive uncertainty (gharar). Without an official screening or certification, io.net does not have a confirmed halal status.

Future Outlook

What to watch

• Supply scale and reliability: More verified GPUs across more regions improves cluster quality and reduces wait times.
• Developer experience: Better tools for data movement, checkpointing, and autoscaling can make distributed jobs easier to run.
• IDE adoption: As more workloads pay fees on‑platform, programmatic burns and USD‑targeted supplier payouts can more closely track real demand.
• Ecosystem partners: Links with storage, rendering, and emerging AI frameworks can make io.net a stronger “intelligent stack” for builders. (io.net)

Longer‑term possibilities

If decentralized compute becomes a standard layer in AI pipelines, io.net could serve as a neutral marketplace that balances supply and demand across many providers. By focusing on orchestration, secure networking, and clear incentives, the network aims to make heavy compute as accessible as a web service call—especially for teams priced out of traditional clouds.

Summary

io.net is a decentralized GPU network on Solana that turns spare compute into an on‑demand cloud for AI and machine learning. Its architecture combines strong orchestration tools, a mesh networking backend, and a dedicated IO token that powers payments, rewards, and staking. The economic model includes a fixed 800 million cap, long‑term hourly emissions, a fee structure that favors paying in IO, and programmatic burns linked to actual network revenue—augmented by the Incentive Dynamic Engine to better match incentives with real usage. With backing from well‑known crypto investors and listings on major exchanges, io.net is positioning itself as a practical, developer‑friendly way to access distributed GPU clusters at scale. Its future will depend on growing reliable supply, deepening ecosystem ties, and continuing to refine both its technology and its incentive design. (io.net)