A mission that started off as an institutional-grade quantitative buying and selling system for cryptocurrencies and shares has transitioned to turn out to be a decentralized network sourcing GPU computing energy to serve growing demand for synthetic intelligence (AI) and machine studying (ML) companies.
Io.web has developed a check network that sources GPU computing energy from quite a lot of knowledge facilities, cryptocurrency miners and decentralized storage suppliers. Aggregating GPU computational energy is touted to drastically cut back the price of renting these sources, which have gotten more and more costly as AI and machine studying advance.
Speaking solely to Cointelegraph, CEO and co-founder Ahmad Shadid unpacks particulars of the network that goals to present a decentralized platform for renting computing energy at a fraction of the price of centralized options that at present exist.
Related: Future of payments: Visa to invest $100M in generative AI
Shadid explains how the mission was conceived in late 2022 throughout a Solana hackathon. Io.web was growing a quantitative buying and selling platform that relied on GPU computing energy for its high-frequency operations however was hamstrung by the exorbitant prices of renting GPU computing capability.
The group unpacks the problem of renting high-performance GPU {hardware} in its core documentation, with the value of renting a single Nvidia A100 averaging round $80 per day per card. Needing greater than 50 of those playing cards to function 25 days a month would value greater than $100,000.
An answer was discovered within the discovery of Ray.io, an open-source library that OpenAI used to distribute ChatGPT coaching throughout over 300,000 CPUs and GPUs. The library streamlined the mission’s infrastructure, with its backend developed within the quick area of two months.
Shadid demoed Io.web’s working testnet on the AI-focused Ray Summit in September 2023, highlighting how the mission aggregates computing energy, which is served to GPU customers as clusters to meet particular AI or machine studying use circumstances.
“Not only does this model allow Io.net to provision GPU compute up to 90% cheaper than incumbent suppliers but it allows for virtually unlimited computing power.”
The decentralized network is ready to leverage Solana’s blockchain to ship SOL (SOL) and USD Coin (USDC) funds to machine studying engineers and miners which might be renting or offering computing energy.
“When ML engineers pay for their clusters, these funds are directed straight to the miners that served in the cluster with their GPUs, with a small network fee being allocated to the Io.net protocol.”
The mission’s roadmap is ready to embrace the launch of a twin native token system that may function IO and IOSD. The token mannequin will reward miners for executing machine studying workloads and sustaining network uptime whereas contemplating the greenback value of electrical energy consumption.
“The IO coin will be freely traded in the crypto market and is the gate to access the compute power, while the IOSD token will serve as a stable credit token algorithmically pegged to 1 USD.”
Shadid informed Cointelegraph that Io.web basically differs from centralized cloud companies like Amazon Web Services (AWS):
“To use an analogy, they’re United Airlines and we’re Kayak; they own planes, whereas we help people book flights.”
The founder added that any companies that require AI computation sometimes use third-party suppliers since they lack the GPUs to deal with all of it in-house. With demand for GPU’s estimated to enhance by 10 instances each 18 months, Hadid stated that there’s typically inadequate capability to meet demand, main to lengthy wait instances and excessive costs.
This is compounded by what he describes as inefficient utilization of information facilities that aren’t optimized for the kind of AI and machine studying work that’s quickly growing:
“There are thousands of independent data centers in the U.S. alone, with an average utilization rate of 12%–18%. As a result, bottlenecks are being created, which is having the knock-on effect of driving up prices for GPU compute.”
The upside is that the typical cryptocurrency miner stands to achieve by renting out their {hardware} to compete with the likes of AWS. Hadid stated that the typical miner utilizing a 40GB A100 makes $0.52 a day, whereas AWS is promoting the identical card for AI computing for $59.78 a day.
“Part of the value proposition of Io.net is, first, we allow participants to be exposed to the AI compute market and resell their GPUs, and for the ML engineers, we are significantly cheaper than AWS.”
Figures shared with Cointelegraph estimate that miners with GPU sources at their disposal might make 1,500% greater than they’d from mining quite a lot of cryptocurrencies.
Magazine: Blockchain detectives: Mt. Gox collapse saw birth of Chainalysis