Aethir’s decentralized Graphics Processing Units (GPUs) network started out as a solution for cloud gaming, but co-founder Mark Rydon says the company is now focused on artificial intelligence — an industry where the increasing demand for in compute power is increasing the cost of GPU and colliding with US – China Geopolitics.
Aethir, founded in 2022, is one of the most talked about protocols in the Decentralized Physical Infrastructure (DePIN) space due to its real-world utility.
DePIN protocols are networks where people work together to offer a service in exchange for a token.
In the past year, space has become extremely hot, with CoinGecko provides DePIN coins a $33 billion market cap. Hundreds of DePINs are emerging for every vertical imaginable from flight tracking data to mapping.
Aethir is one of the most prominent decentralized computing projects, showing that AI can be powered by distributed computing as well as large, centralized data centers.
Let’s travel
GPUs used to be the sole domain of gamers. Companies like Nvidia and AMD have raced to make their GPU chips more powerful so gamers can enjoy their virtual worlds with greater fidelity and resolution.
And then, sometime around 2006Nvidia researchers have discovered that the parallel computing power of GPUs – which makes graphics better and better – is also good for crunching massive amounts of data and can speed up traditional workloads in computing.
This series is brought to you by Consensus Hong Kong. Come and experience the most influential event in Web3 and Digital Assets, Feb.18-20. Register now and save 15% with code CoinDesk15.
But it was 2006 and verticals like AI weren’t on anyone’s radar. Nvidia continued to trade under $1.
Fast forward to now, and Nvidia is challenging Apple for the title of the most valuable company in the world. Over the past few years, GPUs have become the backbone of the AI revolution, powering everything from generative AI models to advanced machine learning algorithms.
Like Nvidia and GPU’s journey, Aethir began by offering idle GPUs from data centers to power cloud gaming platforms previously, as co-founder Mark Rydon explained in an interview with CoinDesk, which decided that the real money is in AI.
“We started Aethir within the context of gaming,” Rydon said. “But we quickly realized that the enterprise-capable GPU cloud we’re building has tremendous relevance to the AI sectors. AI runtime customers will pay more than gamers, and the demand for computing in AI is incredible.”
Unbelievable, really. TrendForce, a market research firm, puts the value of the AI server industry to be $205 billion, which is impressive as they estimate that the value of servers, in total, is just over $300 billion.
Aethir serves as a marketplace for GPU compute, offering idle GPUs to businesses that need on-demand capacity without the overhead of owning or maintaining their own hardware.
For researchers outside of prestigious institutions, this is great. They can get access to computing power, without the huge capital cost of buying their own infrastructure.
“It’s the most expensive computing there’s ever been, and if you’re not good at managing that computing, it’s a huge capital drain,” Rydon said in an interview on the sidelines of Devcon in Bangkok in November. “Our decentralized approach allows us to scale to regions…that need the most computation but can’t afford it.”
Know your customer, know your computer
Given Nvidia’s leadership in the AI space, Chinese researchers are eager to gain access to these ultra-powerful chips that are generations ahead of what’s available domestically.
But geopolitics got in the way. The US Department of Commerce sale prohibited of Nvidia’s most advanced chips, such as the H100, in China to keep Beijing one step ahead.
In August 2024, the The Wall Street Journal reported that Chinese companies are circumventing US export controls by renting decentralized GPU computing power overseas using companies like Aethir.
“We do not accept high performance computing (requests) from places like China. It’s not possible if you have an H100,” said Rydon. “They’re not meant to be in China. We geofence and don’t allow that computing onboard.”
For all the push toward decentralization in Web3, Rydon explains that Aethir is not completely permissionless.
“It was necessary for the $100 million deal to happen,” Rydon said.
There must be a Web2 layer to monitor service agreements that define performance and reliability standards for enterprise clients, and Know Your Customer (KYC) processes, which verify the identities of participants to ensure regulatory compliance.
“I don’t think anyone knows how to build truly permissionless hardware networks,” he said. “The risk to your bottom line as a business is too high if someone can get on and offboard whenever they want with impunity.”
