1. Project situation
1.1 Business Overview
io.net is a decentralized GPU network designed to provide computing for ML (machine learning). Gain computing power by assembling more than 1 million GPUs from independent data centers, cryptocurrency miners, and projects like Filecoin or Render.
Its goal is to combine 1 million GPUs into DePIN (Decentralized Physical Infrastructure Network) to create an enterprise-grade, decentralized distributed computing network that provides AI engineers with cheaper, more accessible, and more flexible network computing resources by aggregating the world's idle network computing resources (currently mainly GPUs).
For users, it acts as a decentralized, global marketplace of idle GPU resources, where AI engineers or teams can customize and purchase the GPU computing services they need according to their needs.
1.2 Team Background
is the founder and CEO, and previously was a quantitative systems engineer at WhalesTrader.
Garrison Yangis Chief Strategy Officer and Chief Marketing Officer, and previously Vice President of Growth & Strategy at Ava Labs.
Tory Greenis the Chief Operating Officer, and previously was the Chief Operating Officer of Hum Capital and Director of Corporate Development & Strategy at Fox Mobile Group.
Angela YiA graduate of Harvard University, he is the Vice President of Business Development, where he plans and executes key strategies for sales, partnerships, and supplier management.
In 2020, when Ahmad Shadid built a GPU computing network for Dark Tick, a machine learning quantitative trading company, the high cost of GPU services from cloud service providers became a problem because the trading strategy was close to high-frequency trading, which required a lot of computing power.
The huge demand for computing power and the high cost they faced led them to decide to go for decentralized distributed computing resources, which they subsequently gained traction at the Austin Solana Hacker House. Therefore, io.net belong to the team to start from the pain points they face, propose solutions, and implement and expand their business.
1.3 Products/Technologies
Availability is limited, using cloud services such as AWS, GCP, or Azure to access hardware often takes weeks, and popular GPU models on the market are often unavailable.
There are few options, such as GPU hardware, location, security level, latency, and so on.
Higher cost: Getting a good GPU is expensive, costing hundreds of thousands of dollars per month for training and inference.
Solution:By aggregating underutilized GPUs (e.g., independent data centers, crypto miners, and crypto projects like Filecoin and Render), and consolidating these resources into DePIN, engineers have access to a lot of computing power in the system. It allows ML teams to build inference and model service workflows across a distributed GPU network and leverage distributed computing libraries to orchestrate and batch training jobs so that data and model parallelism can be parallelized across many distributed devices.
In addition, io.net leverages a distributed computing library with advanced hyperparameter tuning to check the best results, optimize scheduling, and simply specify search patterns. It also uses an open-source reinforcement learning library that supports production-grade, highly distributed RL (reinforcement learning) workloads as well as simple APIs.
Product Composition:IO Cloud, which aims to deploy and manage decentralized GPU clusters for on-demand distribution, seamlessly integrates with the IO-SDK to provide a comprehensive solution for scaling AI and Python applications. Provides unlimited computing power while simplifying the deployment and management of GPU/CPU resources.
IO Worker, which provides users with a comprehensive and user-friendly interface to efficiently manage their GPU node operations through an intuitive web application. The range of products includes features related to user account management, computing activity monitoring, real-time data display, temperature and power consumption tracking, installation assistance, wallet management, security measures, and profitability calculations.
IO Explorer, which provides users with comprehensive statistics and visualizations of all aspects of the GPU cloud, allows users to easily monitor, analyze, and understand the complex details of the io.net network in real time, providing full visibility into network activity, important statistics, data points, and reward transactions.
Features:Decentralized computing network: io.net adopts a decentralized computing model to distribute computing resources across the globe, thereby improving computing efficiency and stability.
Low-cost access: io.net Cloud provides lower access costs than traditional centralized services, making compute resources accessible to more machine learning engineers and researchers.
Distributed cloud cluster: The platform provides a distributed cloud cluster, where users can select appropriate computing resources according to their own needs and assign tasks to different nodes for processing.
Supporting machine learning tasks: io.net Cloud focuses on providing machine learning engineers with computing resources that make it easier for them to perform tasks such as model training, data processing, and more.
1.4 Development Roadmap
https://developers.io.net/docs/product-timeline
According to the information published in the io.net white paper, the roadmap of the project products is:
From January to April 2024, V1.0 will be fully released, which is committed to decentralizing the io.net ecosystem, enabling it to achieve self-custody and self-replication.1.5 Financing Information
根据公开新闻信息显示，2024年3月5日，io.net对外宣布完成 3000 万美元 A 轮融资，Hack VC 领投，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等参与。
[1] It is worth noting that after this round of financing, the overall valuation of io.net is $1 billion.2. Market data
2.1 Official Website
Judging from the official website data from January 2024 to March 2024, the total number of visits is 5.212M, the average monthly visit is 1.737M, and the bounce rate is 18.61% (low io.net).
2.2 Media Groups
3. Competitive analysis
3.1 Competitive Landscape
io.net's core business is related to decentralized AI computing power, and its biggest competitors are traditional cloud service providers represented by AWS, Google Cloud, and Microsoft Intelligent Cloud Business (represented by Azure). According to the 2022-2023 Global Computing Power Index Evaluation Report jointly compiled by International Data Corporation (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.
【2】Compare the sales revenue of the world's mainstream cloud computing vendors: in 2023, the sales revenue of AWS cloud services will be 9.08 billion US dollars, the sales revenue of Google Cloud will be 3.37 billion US dollars, and the sales revenue of Microsoft's intelligent cloud business will be 9.68 billion US dollars.
【3】The three have a global market share of about 66%, and the market value of these three giant companies is more than a trillion dollars.
https://www.alluxio.io/blog/maximize-gpu-utilization-for-model-training/
In stark contrast to the high revenue of cloud service providers, how to improve GPU utilization has become a key issue. According to a survey by AI infrastructure, most GPU resources are underutilized - about 53% of people think that 51~70% of GPU resources are underutilized, 25% think that the utilization rate reaches 85%, and only 7% think that the utilization rate is more than 85%.
For io.net, the huge demand for cloud computing and the inefficient use of GPU resources are market opportunities.3.2 Advantage Analysis
https://twitter.com/eli5_defi/status/1768261383576289429
io.net biggest competitive advantage is reflected in the niche advantage or first-mover advantage. According to the official data, the total number of GPU clusters currently owned by io.net is greater than 40K, the total number of CPUs is greater than 5600, the total number of Woker Nodes is greater than 69K, the time to deploy 10,000 GPUs is less than 90s, the price is 90% cheaper than its competitors, and the valuation is $1 billion. io.net not only provides customers with a low price of 1–2 discount compared to centralized cloud service providers and permissionless instant onboarding services, but also provides additional startup incentives for computing power providers through the upcoming IO token, jointly helping to achieve the goal of connecting 1 million GPUs.
In addition, compared to other DePIN computing projects, io.net focuses on GPU computing power, and its GPU network is already more than 100 times ahead of similar projects. io.net is also the first in the blockchain industry to integrate the most advanced ML technology stacks (such as Ray clusters, Kubernetes clusters, and mega clusters) into GPU DePIN projects and put them into large-scale practice, which makes it a leader not only in the number of GPUs, but also in the ability to apply technology and model training.
As io.net continues to grow, if it can increase its GPU capacity to 500,000 network-wide concurrent GPUs that compete with centralized cloud service providers, it will be able to provide services similar to Web 2 at a lower cost, and have the opportunity to gradually establish its core position in the field as decentralized through close partnerships with major DePIN and AI players (including Render Network, Filecoin, Solana, Ritual, etc.). The leader and settlement layer of the GPU network bring vitality to the entire Web 3xAI ecosystem.
3.3 Risks and Issues
io.net is an emerging computing resource integration and distribution platform that is deeply integrated with Web3, and the business involved is highly overlapping with traditional cloud service vendors, which makes it face location risks and obstacles in terms of technology and market.
Technical Security Risksio.net As an emerging platform, it has not experienced large-scale application testing, nor has it demonstrated the ability to prevent and respond to malicious attacks. In the face of the access, distribution and management of a huge amount of computing resources, there is no corresponding experience or practical verification, and problems such as compatibility, robustness, and security common in technical products are prone to occur. And if something goes wrong, it's likely to be fatal to io.net, because customers are more concerned about their safety and stability and are not willing to pay for them.
Slow market expansionio.net highly overlap with traditional cloud service providers, which makes it have to compete with traditional AWS, Google Cloud, Alicloud and other direct competition, and even direct competition with second- or third-tier service providers, although io.net has a more favorable cost, but its service system and market system for B-class customers have just begun, which is very different from the existing Web3 industry market operation, so at present, its progress in market expansion is not ideal, which is likely to directly affect its project valuation and token market value performance.
Latest security incidentsOn April 25, io.net founder and CEO Ahmad Shadid tweeted that the io.net metadata API suffered a security incident, and an attacker exploited the accessible mapping of user IDs to device IDs, resulting in unauthorized metadata being updated, and this vulnerability did not affect GPU access, but did affect the metadata displayed to users on the frontend. io.net does not collect any PII and does not disclose sensitive user or device data.
Shadid says the io.net system is designed to be self-healing, constantly updating each device to help recover metadata for any erroneous changes. In light of this incident, io.net accelerated the deployment of OKTA's user-level authentication integration, which will be completed within the next 6 hours. In addition, io.net has launched Auth0 Token for user authentication and prevents unauthorized metadata changes. During the database recovery, users will be temporarily unable to log in. All uptime records are not affected, and this does not affect the vendor's calculated rewards.
4. Token valuation
4.1 Token Model
The io.net tokenomics model will have an initial supply of 500 million IO at genesis divided into five categories: Seed Investors (12.5%), Series A Investors (10.2%), Core Contributors (11.3%), R&D & 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.
Rewards are based on a deflationary model, starting at 8% in the first year and decreasing by 1.02% per month (approximately 12% per year) until the 800 million IO cap is reached. As rewards are distributed, the share of early supporters and core contributors will continue to decrease, and the community's share will grow to 50% once all rewards have been distributed. 【4】
Its token functions include allocating incentives to IO Workers, rewarding AI and ML deployment teams for continuous use of the network, balancing some demand and supply, pricing IO Worker compute units, and community governance.
io.net In order to avoid payment problems caused by the fluctuation of the IO currency price, the stablecoin IOSD was specially developed, which is pegged to the US dollar. 1IOSD is always equal to 1 USD. IOSD can only be obtained by destroying IOs. In addition, io.net is considering some mechanisms to improve network functionality. For example, IO Workers might be allowed to increase the probability of being rented by staking native assets. In this case, the more assets they invest, the more likely they are to be selected. In addition, AI engineers staking native assets can prioritize high-demand GPUs.
4.2 Token Mechanism
IO tokens are mainly used for both the demand side and the supply side, for the demand side, each compute job is priced in USD, and the network will hold the payment until the job is completed. Once the node operator configures its reward share in USD and tokens, all USD amounts will be distributed directly to the node operator, while the share allocated to the token will be used to burn IO coins. All IO coins minted as computational rewards during that period are then distributed to users based on the dollar value of their coupon tokens (calculating points).
For the supply side, this includes availability incentives and compute rewards. Among them, the compute reward is that for jobs submitted to the network, the user can choose the time preference "Duration of Deploying the Cluster in Hours" and receive a cost estimate from the io.net pricing oracle. In terms of availability rewards, the network will randomly submit small test jobs to assess which nodes run regularly and are well receptive to jobs from the demand side.
It is worth mentioning that both the supply side and the demand side have a reputation system that accumulates points based on computing performance and participation in the network to obtain rewards or discounts.
In addition, io.net also sets up ecological growth mechanisms, including staking, referral rewards, and network fees. IO coin holders can choose to stake their token, IO, to node operators or users. Once staking, stakers will receive 1–3% of all rewards earned by participants. Users can also invite new network participants to join and share a portion of the new participants' future revenue. Network fees are set at 5%.
4.3 Valuation Analysis
At present, we can't get accurate revenue data for the projects in the track, so we can't accurately estimate it, so we mainly compare it with Render, which is also an AI+DePIN project in io.net, for your 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 project of decentralized GPU rendering solutions in the AI+Web3 track, with a total GPU resource of 11,946 and a current market value of $3 billion (FDV of $5 billion), while io.net GPU resources are 461772, which is 38 times that of Render, and is currently valued at 1 billion. For the io.net and Render projects, the core key capabilities of both are decentralized GPU computing power, so from the perspective of GPU supply as the core comparison dimension, the market value of io.net is likely to exceed render, or at least comparable.
https://stats.renderfoundation.com/
Render network2022年的Frames Rendered是9,420,335,GMV为2,457,134美元,目前,Render Network的Frames Rendered是31,643,819,由此推算整个GMV大概在8,253,751美元。
Control io.net 4-month GMV is 400,000, assuming io.net According to the average growth rate of 4-month GMV of 400,000, the 12-month GMV is 1200,000, if io.net to reach the current GMV of Render Network, there is still 6.8 times the room for growth, and now io.net is valued at $1 billion, combined with the above analysis, the market value of io.net in the bull market cycle is expected to reach more than $5 billion.
5. Summary
The advent of io.net fills a gap in the field of decentralized computing, providing users with a novel and promising way to computation. With the continuous development of fields such as artificial intelligence and machine learning, the demand for computing resources is also increasing, so io.net has high market potential and value.
On the other hand, although the market has priced a high valuation of io.net $1 billion, its products have not been tested by the market, there are uncertain risks in terms of technology, and whether it can effectively match its supply and demand is also a key variable that determines whether its subsequent market value can reach new highs. Judging from the current situation, the results of the io.net platform on the supply side have been initially revealed, but the demand side has not been fully developed, resulting in the overall GPU resources of the platform not being fully utilized, and how to more effectively mobilize the demand for GPU resources is a challenge that the team has to face.
If io.net can complete the rapid access to market demand, and does not encounter or encounter major risks and technical problems in the process of operation, with its AI+DePIN physical business attributes, its overall business will start a growth flywheel and become the most eye-catching project product in the Web3 field, which also means that io.net will be a high-quality investment target of the branch, let us continue to follow up and carefully verify.
References:
【1】https://www.coincarp.com/fundraising/ionet-series-a/
【2】https://medium.com/ybbcapital/promising-sector-preview-the-decentralized-computing-power-market-part-i-368c0621021a
【3】https://www.crn.com/news/cloud/2024/aws-vs-microsoft-vs-google-cloud-earnings-q4-2023-face-off?page=2
【4】https://www.chaincatcher.com/article/2120813
All views above are for informational purposes only and are not intended as investment advice. If you have any objections, please feel free to contact us for correction.
Follow and join the MIIX Captial community for more cutting-edge information:Twitter ：https://twitter.com/MIIXCapital_CN；
Telegram ：https://t.me/MIIXCapitalcn；
加入MIIX Capital团队:[email protected]
Position: Investment Research Analyst / Operations Manager / Visual Designer