State-of-the-art AI requires massive amounts of computing power, and only the largest companies can compete. Wouldn’t it be great if smaller outfits could challenge tech behemoths with their own AI algorithms? That’s what Pluralis Research thinks, and it’s one of a handful of startups that believes blockchains—essentially decentralized cloud computing networks akin to Amazon Web Services—may be the answer. 

On Wednesday, Pluralis announced that it had secured $7.6 million in pre-seed and seed funding, led by the venture capital firms CoinFund and Union Square Ventures. Topology, Variant, Eden Block, and Bodhi Ventures also participated in the round, along with prominent crypto investor Balaji Srinivasan and Clem Delangue, cofounder of the popular AI platform HuggingFace.

The raise was for equity, with a warrant for future cryptocurrency—if Pluralis decides to launch one, Alexander Long, founder and CEO, told Fortune.

For now Pluralis doesn’t have a product, meaning investors are primarily betting on Long and the veteran team of AI researchers he’s assembled. “I raised the round on, ‘This team is the right team to try and tackle this problem’,” Long said. “No one else is trying. We think we can do it.”

The Pluralis founder has a doctorate in computer science and worked at Amazon for more than three years as an AI engineer. He’s assembled seven other computer scientists, all with doctorates or stints as postdoctoral researchers, to see if it’s possible to build powerful AI algorithms through a decentralized network of servers.

Currently, top-flight AI requires fleets of expensive computers in a warehouse working together to train massive algorithms. Researchers estimated that it took 1,300 megawatt hours of energy, or about as much electricity consumed by 130 U.S. homes in a year, to produce an earlier version of OpenAI’s GPT model, or algorithm. This costs money, which means the only firms that can win the AI arms race have large pockets.

For crypto founders, whose technology is predicated on the ideal of decentralization, the idea that a few corporations could wield so much power is a problem, which is why some AI researchers have begun to think about how to create, or train, powerful AI algorithms through a decentralized network of servers. Gensyn, Prime Intellect, and a handful of firms are already exploring the possibility.

Pluralis’s approach is different, Long said. Most attempts at training AI through a network of decentralized computers require those computers to download the entire model. If the servers are small, that puts a ceiling on the size—and power—of a model. Long wants to research whether it’s possible to train portions of a model, rather than the whole model itself, on one computer. “If you can make the problem precise enough, it often leads to immediate ways you can start to solve it,” he said.

Investors believe Long, who has already started his research, may be on his way, and they have enough faith in him and his team that they’re willing to make a bet—even if it’s a longshot. “If this works out,” Jake Brukhman, founder and CEO of CoinFund, proclaimed to Fortune, ”this is going to change the world.” 

This story was originally featured on Fortune.com

Source link