Barry Silbert isn’t done yet. The billionaire entrepreneur first made his mark in finance when, at age 17, he became the youngest person in the U.S. to obtain a stockbroker license. The Maryland native went on to become a Wall Street trader before launching alternative asset platform Second Market, which he sold to NASDAQ. Silbert then hit it big with Bitcoin, buying a hoard when the price was $11 in 2012, and building the crypto conglomerate known as Digital Currency Group. On Wednesday, he announced his next big project: Yuma, a subsidiary that aspires to compete with the likes of Google and OpenAI in the field of artificial intelligence.
The twist is that Yuma is going all in on a decentralized version of AI—the idea of distributing the powerful technology across a loose network of autonomous contributors instead of relying on a giant tech company to provide the service.
Speaking with Fortune, Silbert likened decentralized AI to the world wide web, which in the 1990s supplanted the “walled garden” version of the Internet run by a handful of tech firms. It’s unclear if a decentralized model of AI can hold its own in an industry where the leading firms depend on massive amounts of data, high priced chips and computing power. But Silbert says he is convinced a permissionless version of AI is better—so much so that he is becoming a hands on CEO for the first time in four years to lead Yuma.
AI and blockchain
Yuma’s decentralized AI ambitions revolve around a blockchain project called Bittensor, which launched in 2021 and offers tokens as incentives to spur people to contribute to a network of AI services. Launched in 2019 by a former Google engineer, Bittensor is not widely known but has attracted the support of wealthy investors including Silbert and venture capitalist Olaf Carlson-Wee, who have been buying up its token known as TAO.
Acknowledging that both tokens and AI have been popular fodder for hucksters, Silbert says Yuma is downplaying the crypto angles as “blockchain scares people away.” He says Yuma’s focus will instead be on helping to build a network of decentralized intelligence and computing services in the form of what Bittensor calls “subnets.” There are akin to applications and Yuma is currently supporting around 60 of them, but Silbert envisions there will soon be thousands.
Still, crypto is very much part of the equation as Bittensor and Yuma are counting on TAO tokens to be the incentives that persuade people to contribute to the decentralized AI network.
Like Bitcoin, the TAO tokens are mined using electricity and will become scarcer overtime with the overall supply capped at 21 million. Currently, the market cap of TAO is around $3.5 billion, which makes it the 34th most popular cryptocurrency, far behind the likes of Ethereum, which is 100 times the size.
For now, the Bittensor network is still in an early stage of development so there is little in the way of everyday AI applications for mainstream users. There is also the question of whether, when these applications do arrive, they will be able to overcome the complexity and clunky user interfaces that have been the hallmark of both crypto and decentralized projects.
Silbert says he is confident that it will not take long for developers to push those complexities to the background and build interfaces where users are unaware they are even using a Bittensor service in the first place.
Meanwhile, Michael Casey, an author and journalist who chairs a group called the Decentralized AI Society, says that the solution to the user design challenge will be supplied by AI itself. He points to the burgeoning world of AI agents and predicts that it will soon be possible for users to rely on those agents to deal with all sorts of finicky applications—including decentralized AI.
Technical challenges facing decentralized AI
Jeff Wilser, who hosts the podcast AI-Curious, says he is intrigued by decentralized AI, and the prospect of creating access to a form of artificial intelligence that is not controlled by large tech companies. But he also points out some obvious challenges: OpenAI and Google possess massive amounts of capital to develop the computing power that a successful AI project requires, and it’s not clear a decentralized version will be able to muster the same resources.
The challenges entails not only purchasing custom chips but building centralized data centers where processing facilities are close together—a concept known as colocation that is key to AI efficiency. That is something a decentralized competitor will struggle to replicate, though in time clusters of services may emerge in close proximity to one another. At the same time, there is plenty of spare computing power lying around and so decentralized AI may be able to get a foothold in areas of the industry, such as training data sets, where speed isn’t of the essence.
Yuma’s vision of a decentralized AI competitor powered by a hidden crypto layer may seem far-fetched to some. But skeptics may want to consider another decentralized project that has been a resounding success: Bitcoin, which is distributed across the world and in 15 years has grown to be bigger than all but a handful of major companies.
“Just like the early days of Bitcoin, which fueled the development of a new form of transparent, borderless money, we’re moving from the digital ownership of assets to the decentralized ownership of intelligence,” said Silbert.