MIIX Capital: io.net Project Research Report
io.net is a decentralized GPU network that provides computing resources for machine learning, with a market value of $1 billion, expected to exceed $5 billion during the bull market cycle.1. Project Overview
1.1 Business Summary
io.net is a decentralized GPU network aimed at providing computing power for ML (machine learning). It aggregates over 1 million GPUs from independent data centers, cryptocurrency miners, and projects like Filecoin or Render to obtain computational capabilities.
Its goal is to combine 1 million GPUs into a DePIN (Decentralized Physical Infrastructure Network) to create an enterprise-grade, decentralized distributed computing network. By pooling idle network computing resources globally (currently mainly GPUs), it aims to provide AI engineers with lower-cost, more accessible, and more flexibly adaptable network computing resource services.
For users, it serves as a global marketplace for idle GPU resources, allowing AI engineers or teams to customize and purchase the GPU computing services they need according to their requirements.
1.2 Team Background
Ahmad Shadid is the founder and CEO, previously a quantitative system engineer at WhalesTrader.
Garrison Yang is the Chief Strategy Officer and Chief Marketing Officer, previously the Vice President of Growth and Strategy at Ava Labs.
Tory Green is the Chief Operating Officer, previously the Chief Operating Officer at Hum Capital and Director of Corporate Development and Strategy at Fox Mobile Group.
Angela Yi is the Vice President of Business Development, a Harvard University graduate, responsible for planning and executing key strategies such as sales, partnerships, and vendor management.
In 2020, Ahmad Shadid built a GPU computing network for the machine learning quantitative trading company Dark Tick. Due to the trading strategy being close to high-frequency trading, a large amount of computing power was required, and the high costs of GPU services from cloud service providers became a challenge for them.
The enormous demand for computing power and the high costs they faced prompted them to pursue decentralized distributed computing resources, which gained attention at the Austin Solana Hacker House. Thus, io.net emerged as a solution to the pain points faced by the team, proposing a solution and expanding the business.
1.3 Product/Technology
Problems faced by market users:
Limited availability: Accessing hardware through cloud services like AWS, GCP, or Azure often takes weeks, and popular GPU models are usually unavailable.
Limited choices: Users have almost no options regarding GPU hardware, location, security levels, latency, etc.
High costs: Obtaining quality GPUs is very expensive, costing hundreds of thousands of dollars per month for training and inference.
Solutions:
By aggregating underutilized GPUs (e.g., from independent data centers, crypto miners, and projects like Filecoin and Render), these resources are integrated into DePIN, allowing engineers to access substantial computing power within the system. It enables ML teams to build inference and model service workflows across a distributed GPU network and utilize distributed computing libraries to orchestrate and batch training jobs, allowing for parallelization of data and model workloads across many distributed devices.
Additionally, io.net employs a distributed computing library with advanced hyperparameter tuning to check for optimal results, optimize scheduling, and simply specify search patterns. It also uses an open-source reinforcement learning library that supports production-grade, highly distributed RL workloads and a simple API.
Product Components:
IO Cloud: Aims to deploy and manage on-demand decentralized GPU clusters, seamlessly integrated with IO-SDK, providing a comprehensive solution for scaling AI and Python applications. It offers unlimited computing power while simplifying the deployment and management of GPU/CPU resources.
IO Worker: Provides users with a comprehensive and user-friendly interface to efficiently manage their GPU node operations through an intuitive web application. The product includes features related to user account management, monitoring computing activities, real-time data display, temperature and power consumption tracking, installation assistance, wallet management, security measures, and profitability calculations.
IO Explorer: Primarily offers users comprehensive statistics and visualizations of various aspects of the GPU cloud, enabling them to easily monitor, analyze, and understand the complex details of the io.net network, providing full visibility into network activities, key statistics, data points, and reward transactions.
Product Features:
Decentralized computing network: io.net adopts a decentralized computing model, distributing computing resources globally, thereby improving computational efficiency and stability.
Low-cost access: Compared to traditional centralized services, io.net Cloud offers lower access costs, enabling more machine learning engineers and researchers to obtain computing resources.
Distributed cloud clusters: The platform provides a distributed cloud cluster, allowing users to choose suitable computing resources based on their needs and assign tasks to different nodes for processing.
Support for machine learning tasks: io.net Cloud focuses on providing computing resources for machine learning engineers, making it easier for them to perform tasks such as model training and data processing.
1.4 Development Roadmap
https://developers.io.net/docs/product-timeline
According to the information disclosed in the io.net white paper, the product roadmap is: From January to April 2024, V1.0 will be fully released, focusing on decentralizing the io.net ecosystem to enable self-hosting and self-replication.
1.5 Financing Information
According to publicly available news, on March 5, 2024, io.net announced the completion of a $30 million Series A financing round, led by Hack VC, with participation from Multicoin Capital, 6th Man Ventures, M13, Delphi Digital, Solana Labs, Aptos Labs, Foresight Ventures, Longhash, SevenX, ArkStream, Animoca Brands, Continue Capital, MH Ventures, Sandbox Games, etc. It is worth noting that after this round of financing, io.net's overall valuation reached $1 billion.
2. Market Data
2.1 Official Website
From the official website data from January 2024 to March 2024, the total visits were 5.212 million, with an average monthly visit of 1.737 million and a bounce rate of 18.61% (relatively low). User access data from various regions is relatively uniform, with direct and search access accounting for over 80%, which may indicate that the proportion of dirty data in the visiting user data is low, and they have a basic understanding of io.net and are willing to learn more and interact on the website.
2.2 Social Media Community
3. Competitive Analysis
3.1 Competitive Landscape
The core business of io.net is related to decentralized AI computing power, and its biggest competitors are traditional cloud service providers represented by AWS, Google Cloud, and Microsoft Azure. According to the "2022-2023 Global Computing Power Index Assessment Report" jointly compiled by IDC, Inspur Information, and Tsinghua University's Global Industry Research Institute, the global AI computing market is expected to grow from $19.5 billion in 2022 to $34.66 billion in 2026.
Comparing the sales revenue of major global cloud computing providers: AWS cloud service sales revenue in 2023 was $9.08 billion, Google Cloud sales revenue was $3.37 billion, and Microsoft Azure sales revenue was $9.68 billion. These three companies account for about 66% of the global market share, and all three giants have a market capitalization of over $1 trillion.
https://www.alluxio.io/blog/maximize-gpu-utilization-for-model-training/
In stark contrast to the high revenues of cloud service providers, how to improve GPU utilization has become a focal issue. According to a survey by AI infrastructure, most GPU resources are underutilized—about 53% believe that 51-70% of GPU resources are underutilized, 25% believe utilization reaches 85%, and only 7% believe utilization exceeds 85%. For io.net, the enormous demand for cloud computing and the insufficient effective utilization of GPU resources present market opportunities.
3.2 Advantage Analysis
https://twitter.com/eli5_defi/status/1768261383576289429
io.net's greatest competitive advantage lies in its ecological niche advantage or first-mover advantage. According to official data: io.net currently has a total GPU cluster of over 40K, a total CPU of over 5600, over 69K Worker Nodes, and can deploy 10,000 GPUs in less than 90 seconds, with prices 90% cheaper than competitors, and a valuation of $1 billion. io.net not only offers customers low prices compared to centralized cloud service providers (1-2 times lower) and instant launch services without licensing but also provides additional startup incentives for computing power providers through the upcoming IO token, collectively working towards the goal of connecting 1 million GPUs.
Additionally, compared to other DePIN computing projects, io.net focuses on GPU computing power, with its GPU network scale leading similar projects by over 100 times. io.net is also the first in the blockchain space to integrate advanced ML technology stacks (such as Ray clusters, Kubernetes clusters, and giant clusters) into GPU DePIN projects and put them into large-scale practice, making it a leader not only in GPU quantity but also in technical application and model training capabilities.
As io.net continues to develop, if it can increase GPU capacity to compete with centralized cloud service providers with 500,000 concurrent GPUs, it can provide services similar to Web 2 at lower costs and gradually establish its core position in the field through close collaborations with major DePIN and AI players (including Render Network, Filecoin, Solana, Ritual, etc.), becoming a leader and settlement layer in the decentralized GPU network, bringing vitality to the entire Web 3xAI ecosystem.
3.3 Risks and Issues
io.net is an emerging platform deeply integrated with Web3 for the integration and distribution of computing resources, and its business overlaps significantly with traditional cloud service providers, which poses risks and obstacles in both technology and market positioning.
Technical security risks: As an emerging platform, io.net has not undergone large-scale application testing and has not demonstrated the ability to prevent and respond to malicious attacks. With the massive influx, distribution, and management of computing power resources, it lacks corresponding experience or practical verification, making it prone to common technical product issues such as compatibility, robustness, and security. Moreover, if problems arise, they could be fatal for io.net, as customers care more about their security and stability and are unwilling to pay for these issues.
Slow market expansion: io.net's high overlap with traditional cloud service providers forces it to compete directly with traditional AWS, Google Cloud, and Alibaba Cloud, and even with second- or third-tier service providers. Although io.net offers more favorable costs, its service and market systems targeting B-class customers are just beginning, which differs significantly from the existing market operations in the Web3 industry. Therefore, its current progress in market expansion is not ideal, which could directly affect its project valuation and token market performance.
Latest Security Incident
On April 25, io.net founder and CEO Ahmad Shadid tweeted that the io.net metadata API encountered a security incident, where attackers exploited the accessible mapping from user ID to device ID, resulting in unauthorized updates to metadata. This vulnerability did not affect GPU access but did impact the metadata displayed to users on the front end. io.net does not collect any PII and does not leak sensitive user or device data.
Shadid stated that the io.net system design allows for self-repair, continuously updating each device to help restore any erroneous changes to metadata. In light of this incident, io.net expedited the deployment of user-level identity verification integration with OKTA, which will be completed within the next 6 hours. Additionally, io.net launched Auth0 Token for user verification to prevent unauthorized metadata changes. During the database recovery period, users will temporarily be unable to log in. All uptime records remain unaffected, and this will not impact vendor computing rewards.
4. Token Valuation
4.1 Token Model
The io.net token economic model will have an initial supply of 500 million IO tokens at genesis, divided into five categories: seed investors (12.5%), Series A investors (10.2%), core contributors (11.3%), R&D and ecosystem (16%), and community (50%). As IO is issued to incentivize network growth and adoption, it will grow to a fixed maximum supply of 800 million over 20 years.
The reward adoption follows a deflationary model, starting at 8% in the first year and decreasing by 1.02% each month (approximately 12% annually) until reaching the 800 million IO cap. As rewards are distributed, the shares of early supporters and core contributors will continue to decrease, and after all reward distributions are completed, the community's share will grow to 50%.
Its token functions include providing allocation incentives for IO Workers, rewarding AI and ML deployment teams for continued use of the network, balancing some demand and supply, pricing for IO Worker computing units, and community governance.
To avoid payment issues caused by fluctuations in the IO token price, a stablecoin IOSD has been specially developed, pegged to the US dollar. 1 IOSD is always equal to 1 dollar. IOSD can only be obtained by burning IO tokens. Additionally, io.net is considering mechanisms to improve network functionality. For example, it may allow IO Workers to increase their chances of being rented by staking native assets. In this case, the more assets they stake, the greater their chances of being selected. Furthermore, AI engineers staking native assets can have priority access to high-demand GPUs.
4.2 Token Mechanism
The IO token is primarily used for two major groups: demand-side and supply-side. For the demand side, each computing job is priced in dollars, and the network will retain payment until the job is completed. Once node operators configure their reward shares in dollars and tokens, all dollar amounts will be directly allocated to node operators, while the shares allocated to tokens will be used to burn IO tokens. Then, all IO tokens minted as computing rewards during that period will be distributed to users based on the dollar value of their coupon tokens (computing credits).
For the supply side, it includes availability rewards and computing rewards. Among them, computing rewards are for jobs submitted to the network, where users can choose their time preference for "the duration of cluster deployment in hours" and receive cost estimates from the io.net pricing oracle. Regarding availability rewards, the network will randomly submit small test jobs to assess which nodes operate regularly and can accept jobs from the demand side well.
It is worth mentioning that both supply-side and demand-side have a reputation system in place, accumulating scores based on computing performance and participation in the network to earn rewards or discounts.
In addition, io.net has set up ecological growth mechanisms, including staking, invitation rewards, and network fees. IO token holders can choose to stake their IO tokens to node operators or users. Once staked, stakers will receive 1-3% of all rewards earned by participants. Users can also invite new network participants and share a portion of the future income of new participants. A network fee of 5% is also set.
4.3 Valuation Analysis
Currently, we cannot obtain accurate revenue data for projects in the same track, so we cannot conduct precise valuations. Here, we mainly compare io.net with Render, another AI + DePIN project, for reference.
https://x.com/ionet/status/1777397552591294797
https://globalcoinresearch.com/2023/04/26/render-network-scaling-rendering-for-the-future/
As shown in the figure, Render Network is currently the leading decentralized GPU rendering solution in the AI + Web3 track, with a total GPU resource of 11,946 and a market capitalization of $3 billion (FDV $5 billion); while io.net has a total GPU resource of 461,772, which is 38 times that of Render, with a current valuation of $1 billion. For both io.net and Render projects, the core key capability is decentralized GPU computing power. Therefore, from the perspective of GPU supply as a core comparison dimension, io.net's market capitalization is likely to exceed that of Render, or at least be comparable.
https://stats.renderfoundation.com/
In 2022, Render Network rendered 9,420,335 frames, with a GMV of $2,457,134. Currently, Render Network has rendered 31,643,819 frames, estimating the total GMV to be approximately $8,253,751.
In contrast, io.net's GMV over four months is $400,000. Assuming io.net maintains a steady growth rate of $400,000 GMV over four months, its 12-month GMV would be $1,200,000. If io.net aims to reach Render Network's current GMV, it has 6.8 times growth potential. With io.net's current valuation of $1 billion, based on the above analysis, io.net's market capitalization in a bull market cycle is expected to exceed $5 billion.
5. Conclusion
The emergence of io.net fills a gap in the decentralized computing field, providing users with a novel and promising way of computing. As fields like artificial intelligence and machine learning continue to develop, the demand for computing resources is also increasing, giving io.net high market potential and value.
On the other hand, although the market has given io.net a high valuation of $1 billion, its products have not been tested in the market, and there are uncertainties in technology, as well as whether it can effectively match its supply and demand relationships, which are key variables determining whether its subsequent market value can reach new highs. From the current situation, the achievements on the supply side of the io.net platform have begun to show initial results, but it has not fully exerted its strength on the demand side, leading to the current overall GPU resources of the platform not being fully utilized. How to more effectively stimulate the demand for GPU resources is a challenge that the team must face.
If io.net can rapidly connect to market demand and does not encounter significant risks and technical issues during its operations, its AI + DePIN business attributes will kickstart a growth flywheel, making it one of the most prominent project products in the Web3 field. This also means that io.net will be a high-quality investment target, and we will continue to follow up and observe closely for verification.
Reference Resources
