New York’s CoinFund and Union Square Ventures have led a $7.6 million seed funding round for Pluralis Research, an Australian startup developing a blockchain-based AI platform with global access.
Los Angeles-based Bodhi Ventures, Miami Beach, Fla.-based Topology, New York’s Variant and London’s Eden Block also participated, as did crypto investor Balaji Srinivasan and Clem Delangue, co-founder of the AI platform HuggingFace.
Founded by Alexander Long, a former Amazon engineer, Melbourne-based Pluralis is building a decentralized AI platform that he envisions could be used like Amazon Web Services across the world.
‘AI For All’
“Pluralis paves the way to true collective ownership at the model layer, by training foundation models that are split across geographically separated devices connected only over the Internet,” said Long, who has a Ph. D. in computer science. “It guarantees public access to strong base models, the ability for anyone to participate in the economic benefits of AI, and unlocks open innovation. There may be no technical challenge today that has a larger impact.”
In an interview with Fortune, Long said Pluralis has assembled seven other computer scientists, all with doctorates, in his quest to build powerful AI algorithms through a decentralized network of servers.
Pluralis’ approach differs from other rivals trying to build decentralized AI in the sense it will not require users to download an entire model. Long wants to determine if it is possible to train portions of a model, rather than the whole model, on one computer, Fortune said.
Multiple Trainers
Pluralis is also touting what it calls Protocol Learning, a system that can be trained by multiple contributors without any single entity controlling the full model weights. This framework aims to reshape how foundation models are built, creating a more open and distributed AI ecosystem with sustainable economic incentives, the company said.
“The model ‘lives’ within the protocol and never leaves — it is trainable and usable, but unextractable. This framework, where no one party obtains a full copy of the model weights, allows for value to programmatically flow to contributors, and creates economic incentive to contribute to the network,” the company said.
“If this works out,” Jake Brukhman, founder and CEO of CoinFund, told Fortune, “this is going to change the world.”
