"We believe AGI should be open source, owned and leveraged by humanity, not a handful of corporations." That is how Sachi Kamiya, director of venture and growth at the Sentient Foundation, framed the bet in written answers. On June 24, the nonprofit put money behind the slogan. It committed $42 million through what it calls the Open Source AGI Grant and Investment Program. The number is a wink at "The Hitchhiker's Guide to the Galaxy," in which 42 is the answer to life, the universe and everything.
The project carries real pedigree. Sentient was founded in January 2024 by Sandeep Nailwal, the Polygon co-founder, alongside Princeton professor Pramod Viswanath and Himanshu Tyagi of the Indian Institute of Science. Months later, Sentient Labs, the research arm, raised an $85 million seed round co-led by Peter Thiel's Founders Fund, Pantera Capital and Framework Ventures. The new grant program funds outside builders, in two tracks, one of non-dilutive grants and one of startup investment.
The thesis rests on a claim that would have sounded absurd a year ago. "Most people in the US default to ChatGPT or Claude, assuming they're categorically better," Kamiya wrote, but "open source counterparts are now only 3-6 months behind." Her ambition for Sentient is blunt: "We want to become the Linux Foundation of AI," she wrote, a neutral institution that funds the people closing that gap.
The case for keeping the most powerful models out of corporate hands has a louder champion in Ben Goertzel, the founder of SingularityNET, who has argued the point for two decades. "I'm pretty adamant that the core AGI code doing the thinking should be free and open source," Goertzel said on the On The Margin podcast. "I don't want to give up control of what I'm doing to venture capital firms." His warning is geopolitical, not just commercial. He said he would rather see "an open and decentralized AI" than "an AGI arms race where just a few big countries are the only ones who have AGI and they're using it to try to blow each other up."
Goertzel also flags the catch that money alone does not solve. Open code is not the same as open access if running the model is ruinously expensive. Once a model needs a hyperscale server farm to run, he said on a podcast, "the fact that the code is open doesn't help that much." That is the gap Sentient says its grants and compute are meant to fill.
Here is where crypto enters a story that is ostensibly about AI, an intersection investors have been chasing for the past year. "Crypto and AI are two sides of the same coin," Kamiya wrote. Traditional bank rails are too slow and too permissioned for software to use, she argued, so "blockchain is the transaction layer for AI agents." It is a self-serving claim from a project with a token, but it is not hers alone.
Yat Siu, the co-founder of Animoca Brands, makes the same argument in starker terms. "Is JP Morgan gonna open an account for an agent? Probably not gonna happen," Siu said on the a podcast. "They have a wallet." He expects the population of these agents to dwarf the human one. "You're gonna have anywhere from 50 to 100 billion agents minimum," he said, each one needing a way to pay and get paid that no bank will provision.
Some of that plumbing already exists. Marc Boiron, chief executive of Polygon Labs, said on the a podcast that the rails are further along than outsiders assume. "Most people don't realize yet. Polygon is just a payments focused chain," he said, describing a network already moving stablecoins for consumer apps. Nitya Subramanian, founder of the wallet-infrastructure startup Para, said on the On The Margin podcast that "wallets are ultimately the authorization and control flow layer of anything that's happening on chain." She offered a homely example of what an agent budget looks like in practice, the kind of programmable stablecoin card Visa is already chasing: "I could create a stable coin backed card and give it $200 a week and just have it buy Chipotle."
The vision has an obvious failure mode, and not everyone selling it pretends otherwise. Varun Kabra of Concordium said on the On The Margin podcast that the next phase, in which agents act on their owners' behalf, is already underway, with "AI agents start transacting on your behalf." The problem is on the other side of the trade. A merchant has no way to verify that a real, accountable human is behind the transaction, Kabra said, which "could open a door to fraud, bots acting as humans, agents operating with no accountability." Open-source models do not close that hole on their own.
Sentient's harder problem is money, the same one that has sunk open-source business models before it. Kamiya called monetization "the existential question for open source." Sentient's answer is a framework it calls OML, for open, monetizable and loyal models, paired with a fingerprinting technique that fine-tunes a model to give signature responses so its provenance can be proved even after it is copied. Whether that survives contact with an industry built on closed APIs is unproven.
For now, the $42 million buys Sentient the leadership of a movement that, until recently, had no foundation to call its own. Whether it delivers comes down to the benchmark Kamiya set for herself. Asked what success looks like in five years, she wrote: "The goal is for open source AI to be on par with closed source in performance."