Introduction to Decentralized GPU Rental Platform io.net, Targeting the AI Computing Market
What is the decentralized AI computing resource platform io.net? IO tokens are expected to be launched on Solana in the near future. What are the advantages and challenges of the DePIN project that has attracted market attention? This article will provide a detailed analysis and introduction.
Contents:
Toggle
Introduction to io.net
What problem does io.net want to solve?
Large global GPU market gap
Increasing hardware threshold for AI startups
Design concept of io.net
Supply side: Incentive model to attract market rental of idle computing resources
Demand side: Customers can effectively utilize computing resources through cluster division
Potential risks of io.net
Token economics of io.net
Basic information of IO tokens
Utility of IO tokens
Distribution model of IO tokens
IO token airdrop opportunity
Solana seizes the DePIN narrative
io.net, established at the Solana hackathon in February 2023, is a decentralized GPU distribution infrastructure built on the Solana network. It aims to provide more efficient and flexible computing power for machine learning, making it more scalable for development teams to train model calculations.
In just one year of development, due to the compatibility of the theme with industry trends and the appropriate development of the team, io.net has already received a $30 million investment in March of this year. Investors include Hack VC, Delphi Digital, Solana, Aptos, Foresight Ventures, ArkStream Capital, and other well-known institutions, which have attracted more attention to this DePIN project.
In recent years, due to the rapid development of the AI industry, existing cloud computing services are not sufficient to meet the demand for machine learning and AI-related computing in the market, and global computing resources are gradually becoming inadequate.
AI development teams need to face a seller’s market, which indirectly leads to their dependence on cloud suppliers for GPU hardware specifications, location, security levels, latency, and other options. Buyers have almost no choice, let alone the inability to adopt mainstream GPU models for customized needs.
Existing CPU and GPU computing capabilities and quantities cannot meet the needs of machine learning applications, and considering the longer supply cycle of GPUs, this problem is unlikely to be solved quickly.
Due to the scarcity of computing resources and the increasing competition of AI models, small companies and teams without substantial funding cannot afford to purchase a large number of computing devices and management costs (which may reach hundreds of thousands of dollars per month), resulting in low training efficiency and long-term difficulty in gaining competitiveness.
The market needs a more flexible way to allocate computing resources.
io.net attempts to solve the above problems by utilizing underutilized GPU resources outside of cloud services. These computing resources may come from:
Independent data centers: There are thousands of independent data centers in the United States alone, but the average utilization rate is only 12% to 18%.
Cryptocurrency miners: After Ethereum transitioned to proof-of-stake (POS), some miners have idle capacity.
Consumer-grade GPUs: Consumer-grade GPUs account for 90% of the total supply, but most resources are hidden in consumers’ homes and are not fully utilized.
Overall, the io.net team estimates that these computing resources can provide more than ten times the additional computing resources. Therefore, the team’s goal is to build a service architecture that can effectively utilize these resources.
By using the incentive model of cryptocurrencies, io.net allows users with idle GPU resources to rent out their computing resources. These computing resources will be consolidated in the decentralized network of io.net, allowing the required development teams to access and customize the necessary computing resources freely, quickly, and without permission.
The rental fees will be determined based on GPU specifications, brands, security, compliance certification, etc. However, the team states that through its integration capabilities and unique pricing mechanism, it can be 90% cheaper than traditional cloud suppliers, giving it an advantage in renting computing resources.
According to the io.net team, their service has higher computational efficiency
(Source)
Clients renting computing resources can maximize their own needs and efficiency by choosing existing diverse clusters or creating unique filtering conditions for clusters. For example, they can choose a computing resource cluster that is 100% powered by green energy to meet ESG goals, or choose computing resources in Taiwan to reduce costs and accelerate connection speeds.
Commonly available clusters in io.net
On the other hand, compared to traditional cloud suppliers like AWS, io.net has no contract restrictions and does not require permission. Users can access and deploy clusters within 90 seconds.
The actual adoption of io.net in the mainstream market still remains uncertain. Both computing resource suppliers and rental teams are relatively unfamiliar with this model, and the actual implementation is expected to face certain challenges.
On the other hand, there will also be many short-sighted hunters participating in other non-compliant ways, as io.net recently stated that it will prohibit virtual GPU suppliers from participating. Effectively managing the incentive model to promote services without causing counterproductive effects is a major challenge for the io.net team.
io.net uses blockchain as the settlement network for its incentive layer, which allows it to use token incentive models to create such decentralized business models that integrate idle computing resources. The native token of the io.net project is IO, but it is not currently officially launched, so detailed information and data may change.
Name: IO Coin
Mainnet: Solana (SPL)
Address: Not online, expected to be launched in the near future
Total supply: 800 million tokens
Token for GPU renters:
If these rental development teams use IO to pay for the rental of GPU clusters, they will not be charged a fee (paying with USDC incurs a 2% fee).
Token staking requirement for GPU suppliers:
Computing resource suppliers need to stake a certain amount of IO tokens to receive idle rewards (IO token rewards when not rented). The specific quantity will require more information from the official team.
Token staking:
Anyone holding IO tokens can stake the tokens on the Solana protocol to receive staking rewards.
According to the io.net whitepaper, there is a total of 800 million IO tokens, of which 500 million tokens will be released at launch, and the remaining tokens will be unlocked and distributed to GPU suppliers and token stakers every hour. The release speed will be 8% in the first year, decreasing by 1.02% each month, and it is expected to be distributed in about 20 years.
io.net is expected to have a token buyback and burn mechanism, which means that part of the team’s income will be used to repurchase IO tokens on the market and burn them, indirectly rewarding token holders.
Detailed numbers and models still require further updates from the official team.
Since the team has indicated that they will issue tokens, and the only way to participate at the moment is to rent or lease computing resources, regular users may try renting out their home computer resources to have a chance to participate in io.net’s airdrop activities.
(This article is not investment advice)
Solana has recently received a lot of market attention. In addition to meme coins and NFTs, many in the community believe that given the high efficiency of its network, DePIN may have the opportunity to become Solana’s next growth driver, clearly demonstrating Solana’s efforts in this area.
Recommended reading:
DePIN | Messari compiles industry development status and project introductions
Reason for recommendation: This article takes a higher-level view of the overall development status and trends of the DePIN industry, and lists various applications and projects in different fields, providing readers with a more comprehensive understanding of this emerging industry. The article also points out the advantages of Solana in developing DePIN.
In addition to nurturing io.net, Solana has attracted many projects to join its ecosystem, including Helium, Render Network, Hivemapper, and other leading applications.
(Decentralized network Helium Mobile launches unlimited 5G plan in the US, cheaper than Chunghwa Telecom)
(AI veteran project Render Network (RNDR) joins the Solana network)
Whether it is self-cultivated teams in the ecosystem or business development efforts to attract existing projects, Solana undoubtedly has a place in the DePIN field.
AI
DePIN
io.net
Solana
Decentralized computing
Further reading:
Binance Research 2024 Report | Key trends in blockchain and cryptocurrency
DePIN | Messari compiles industry development status and project introductions
