Io.net: Revolutionizing GPU Computing with Decentralized Network for AI and Machine Learning
Summary:
Io.net, initially a quantitative trading system for cryptocurrencies and stocks, is now a decentralized network amassing GPU computing power to cater to the rising demand for AI and machine learning services. The network significantly cuts the cost of outsourcing GPU power by drawing from diverse sources, making it a cost-effective alternative to existing centralized platforms. The network's dual native token system will incentivize miners to carry out machine learning tasks, maintain network uptime, and consume electricity efficiently. Io.net's fundamental difference from services like AWS is explained using an analogy – they own the 'planes', while Io.net assists in 'booking flights'. Miners can earn revenue by leasing their computing power harnessing the high demand for such services.
What began as an enterprise-grade quantitative trading system for cryptocurrencies and stocks has evolved into a decentralized network that pools GPU computing power to meet the growing demand for AI and machine learning services.Io.net has designed a test network that pools GPU computing power from various data centers, cryptocurrency miners, and decentralized storage suppliers. This pool can significantly decrease the cost of these increasingly expensive sources as the demand for AI and machine learning grows. Ahmad Shadid, CEO and co-founder, in an exclusive interview with Cointelegraph, discusses his network's mission to provide a more affordable computing power leasing option than today's centralized alternatives.The concept for the project emerged during a Solana hackathon in late 2022. Io.net was focusing on a high-frequency-operating quantitative trading platform dependent on GPU computing power, but was hindered by the high price of sourcing such power. The io.net platform aims to offer GPU computing resources for AI and machine learning requirements, as an alternative to traditional methods. The team highlights the issue of leasing high-performance GPU hardware and reveals that it costs approximately $80 per day to rent a single NVIDIA A100. Addition of over 50 of these cards needed operation for 25 days each month could run up a bill exceeding $100,000. A solution emerged with the finding of Ray.io, an open-source library used by OpenAI to distribute ChatGPT training across a massive number of CPUs and GPUs. This library streamlined the project, reducing overall backend development time to just two months. In the Ray Summit in September 2023, Shadid demonstrated io.net’s testnet highlighting how the project collates computing power provided to consumers in clusters for specific AI and machine learning applications.“Not only will this model allow io.net to offer GPU computation at nearly 90% less than existing suppliers, but it also promises limitless computing power.”As revealed, the decentralized network will utilize Solana’s blockchain to facilitate SOL and USD Coin (USDC) payments to machine learning engineers and miners involved in providing or leasing computing power. Fees for clusters provided by ML engineers go directly to miners, with a nominal fee allocated to io.net. Looking ahead, the project aims to launch a dual native token system featuring $IO and $IOSD with the goal to incentivize miners to complete machine learning tasks, maintain network uptime, and consume electricity wisely.Shadid declared io.net is unlike centralized cloud services such as Amazon Web Services (AWS) by saying, “they’re similar to United Airlines and us to Kayak; they own airplanes, we assist in booking flights.” Increased demand for AI computation has led to businesses seeking help from third-party providers. Lack of in-house GPUs means there is often insufficient capacity resulting in long wait times and high pricing. There is an inefficient utilization of data centers, as Shadid describes, which are not optimized for AI and machine learning's rapid demands. This inefficiency drives the cost for GPU computation high. Cryptocurrency miners, however, could earn by renting their hardware to compete with giants like AWS. According to Hadid, an average miner utilizing a 40GB A100 makes $0.52 a day, while AWS sells the same card for $59.78 a day for AI computing. The proposition of io.net opens avenues for participants to access and sell their GPUs in the AI compute market wherein the platform is significantly more affordable than AWS. Cointelegraph received data suggesting that miners leasing their GPU resources may earn 1500% more than they would by mining various cryptocurrencies.
Published At
10/11/2023 1:00:00 PM
Disclaimer: Algoine does not endorse any content or product on this page. Readers should conduct their own research before taking any actions related to the asset, company, or any information in this article and assume full responsibility for their decisions. This article should not be considered as investment advice. Our news is prepared with AI support.
Do you suspect this content may be misleading, incomplete, or inappropriate in any way, requiring modification or removal?
We appreciate your report.