Why AGI Must Be Decentralized
Ben Goertzel, often called the 'Godfather of AGI,' made a clear case for decentralized artificial general intelligence in a recent podcast. "I don't want to give up control of what I'm doing to venture capital firms," he said on the On the Margin podcast. "Because I think AGI is too important for that." Goertzel, who popularized the term AGI with an influential book and led the team behind Sophia the robot, has spent decades pursuing machines that can think. His current bet runs counter to most others racing toward the same goal: the most important software humans ever write should not be owned by a single company.
That conviction drives SingularityNET, the blockchain project he runs, and the broader Artificial Superintelligence Alliance formed in 2024 with Fetch.ai and Ocean Protocol (Ocean left in late 2025). While OpenAI and Anthropic raise tens of billions and keep their best models behind closed doors, Goertzel wires his AGI work to a crypto network owned by its users. He argues that open source alone is insufficient: "If the code is open but the data takes a server farm and you need a hyperscaler server farm to use it, the fact that the code is open doesn't help that much." Instead, he wants AGI to run on a decentralized network spanning 50 countries and controlled by 10,000 different people.
Critique of Closed AI Labs
Goertzel is blunt about companies that once shared his open ethos then walked away. He points to the Elon Musk vs. Sam Altman litigation as evidence of OpenAI's rapid pivot: "Sam Altman didn't think open for very long; he rapidly shifted to making it proprietary." He is harsher on Anthropic: "Dario Amodei's stuff was closed from the very beginning." Even Musk is criticized for moving from warning about AI to building his own closed xAI. Goertzel concedes the closed path is easier but insists the open one is possible: "Linux and the internet are proof points that something can be open source, globally decentralized, and still be the foundation for people making a lot of money."
Revenue Model and Agent Economy
The business currently runs on crypto tokens. "We have people running nodes hosting AI processes; we make money from tokenomics," Goertzel says. He plans to shift toward web2 offerings, selling polished AI products with the blockchain hidden from end users. A paid tier comparable to Claude Pro or ChatGPT Pro but "smarter" will launch next year. He will not chase retail chatbots that lose money.
Goertzel envisions an 'agent economy' where AI agents transact on behalf of users. He predicts the next wave of victories will come to small groups orchestrating teams of AI agents. Varun Kabra of Concordium agrees agents will handle payments and subscriptions. Nitya Subramanian of Para warns about the risks of outsourcing purchases. Goertzel's solution is to run this economy on an open network rather than a corporate cloud.
Timeline and Importance
Goertzel still expects human-level AGI by 2029, possibly 2027 or 2030. His worry is inequality: "If there's an understanding gap between people at the top and bottom, it will exacerbate inequality." That's why he insists on broad access. His first test, the Omega Claw agent, will launch within weeks, allowing users to teach personal agents to manage their lives and make money. Whether a crypto network can outcompete companies with tens of billions remains unproven, but Goertzel has offered a distinct alternative: user-owned, decentralized, and open-source AGI.