In a two-story brick office building in Emeryville, neuroscientists are making plans to fit mice with diminutive brain-computer interfaces to record patterns of neural activity while they perform basic tasks like navigating a maze. They hope to build a library of mouse brain states that reliably maps to specific perceptions and actions.
Next comes translation: turning those findings into code—and, eventually, into a new kind of AI system built on the brain’s governing principles. They plan to run the experiments on mice, monkeys and even humans.
If it works, it could become a flywheel: brain experiments inform new AI architectures and those suggest new hypotheses to test. Hovering over all of it is a science fiction-sounding ambition: using brain computer interfaces not just to read minds, but to write them. Researchers talk about “uploading” knowledge into the brain, inserting the image of an apple into someone’s thoughts, or directions to navigate a never-before-seen maze.
It sounds like an idea pulled straight out of a William Gibson book. Jed McCaleb, who founded cryptocurrency projects Ripple and Stellar, is trying to bankroll it into reality.
The Silicon Valley billionaire is committing $1 billion of his cryptocurrency fortune–worth about $3.9 billion by Forbes’ estimate–into building AI systems that achieve artificial general intelligence, or the threshold at which AI systems can perform tasks as well as humans.
“Most effort and research… is going in one particular area—transformers,” McCaleb said. “[AI] would benefit by looking closer at the human brain.”
He's funding the effort through his nonprofit Astera Institute, which has long had an AI focus, but recently zeroed in on brain-inspired approaches, prompting McCaleb’s nine-figure commitment. In doing so, Astera joins a slew of other funded-into-the-billions AI research ventures with no clear commercialization plan, including former OpenAI co-founder Ilya Sutskever’s SSI or Jeff Bezos’ Project Prometheus.
The self-made son of a single mother in Arkansas, McCaleb fits the modern technologist billionaire mold: future-obsessed, capital-rich and impatient with playing it safe. His bets are not small. In addition to Astera Institute, where he has pledged $600 million towards neuroscience on top of the $1 billion AGI effort, he spends several days a week at Vast, his space company in southern California. Vast wants to replace the international space station, and Forbes first reported it has been fundraising at a $2 billion valuation.
“Crypto was, in some sense, a big detour,” McCaleb says. “I’d been wanting to work on AI the whole time, but only really got the chance once I stepped back [from crypto]...I think AI is going to be the most transformative thing that humans ever create. So it’s the most compelling thing to work on.”
McCaleb founded Astera Institute in 2020 with his spouse Seemay Chou, a former University of California, San Francisco biology professor. Last year, they pledged to give away the majority of their wealth via the Giving Pledge, which was created in 2010 by billionaire philanthropists Bill Gates, Melinda French Gates and Warren Buffett in an effort to charitable causes.
On the AGI front, McCaleb’s first big hire is former DeepMind executive Dileep George. George previously co-founded two AI companies: Vicarious AI, which was acquired by Google parent company Alphabet and before that, Numenta, an early neuroscience-focused AI company he launched with PalmPilot creator Jeff Hawkins.
George hopes to grow the lab to 30 researchers this year. Hiring is a priority: while Astera can’t compete with Brobdingnagian compensation packages offered by OpenAI and the big tech companies, it is targeting researchers who are motivated by the mission.
“A philanthropy-supported approach is better at this time because there are core research problems to be solved,” George said. “Startups have to worry about the next fundraise and the next demo that will drive the fundraise—and that’s a distraction.”
Another potential recruiting lever: Astera plans to publish its research openly, echoing early OpenAI—before the industry’s insatiable need for capital and competitive pressures inspired it to abandon it for a secretive, for-profit structure.
Astera’s mission is easy to state, but daunting to execute— study how the brain works, then use those insights to build AI systems that are more brain-like. George hopes this will lead to new AI architectures that are more efficient, transparent and controllable than today’s. When McCaleb and George first met, they bonded over their shared skepticism that scaling transformers – the AI architecture underpinning products like ChatGPT – will produce AGI. They think the field needs to go back to the drawing board.
OpenAI CEO Sam Altman disagrees, telling Forbes in January that he believes AGI will require “a lot of medium-sized breakthroughs” rather than a single sweeping new one.
McCaleb’s critique is simple. “Transformers are probably just doing one aspect… this kind of prediction,” McCaleb said, adding that key elements like planning, decision-making and motivation are still missing.
The broader research tide is moving in a direction similar to Astera’s. Former Meta Chief AI Scientist Yann LeCun’s new AI research lab, AMI, recently raised a $1 billion seed round to advance the work he did at Meta building “world models,” mental representations of the world around us.
McCaleb touts brain-inspired AI as a pathway to safer AI. Today’s frontier models are gargantuan and opaque; understanding what they “know” and why they give the answers they do remains difficult.
“If it works more like the human brain, there’s a better chance we can understand it… rather than being this kind of abstract mathematical thing that ends up being very alien,” McCaleb said.