Thomas Reardon has spent years thinking about how brains work—first at Microsoft, where he led Internet Explorer development, then at CTRL-labs, the neural interface company Meta scooped up in 2019. Now he's convinced the AI industry is doing it all wrong.
His new venture, Flourish, reportedly secured $500 million at a valuation around $2.5 billion, according to WIRED. The New York startup counts Jeff Bezos, Google's venture arm GV, Lux Capital, and Catalio Capital among its backers, WIRED reports. What Flourish is promising in return is ambitious, maybe audacious: artificial intelligence that matches human cognition while sipping power like a human brain does, running on 50 watts or less.
That's roughly a fourteenth of what an Nvidia H100 GPU guzzles under load. For context, training today's frontier models can require server farms pulling megawatts. Reardon, who holds a neuroscience PhD from Columbia, believes the path to truly intelligent machines runs through biological principles, not just bigger compute clusters.
Memory, Learning, and Silicon
The company calls its technology Cortex AI. According to WIRED's June 4 report on the launch, Flourish is tackling "the two most difficult problems facing AI today: power efficiency and continuous learning." One model in development mimics the hippocampus to manage memory functions. Another focuses on enabling continuous learning in mobile environments, where battery constraints actually matter.
Flourish told WIRED it's in talks with a major chip manufacturer about implementing these models directly in hardware, though the company wouldn't disclose which partner. If that materializes, it would mark a notable shift from software-only AI development to purpose-built neuromorphic silicon.
The company operates from a West SoHo building outfitted with its own data center. Unlike most AI labs, Flourish runs wet-lab neuroscience experiments alongside its computational work. The setup reflects Reardon's conviction that understanding biological neural networks isn't optional.
Reunions and Departures
Reardon isn't building this alone. His co-CEO is Robert Williams, who recently left Amazon's S-team after overseeing Alexa and VegaOS. The two other co-founders bring academic credentials: Joshua Vogelstein, an associate professor at Johns Hopkins now on leave, and Vish Sivakumar, who worked in Meta's Neural Interfaces Group alongside Reardon.
The team has attracted some notable outside support. Greg Wayne, a researcher at DeepMind, is devoting a fifth of his time to Flourish. Ben Recht, a professor at UC Berkeley known for his skepticism about AI hype, signed on as an adviser. By late March, the company had pulled together roughly two dozen neuroscientists and AI researchers, WIRED reported, though Flourish's LinkedIn profile showed 14 visible profiles as of early September.
That employee count might seem modest for a company valued north of $2 billion. Then again, Reardon isn't exactly chasing the standard playbook.
The Bezos Factor
Jeff Bezos moved fast. After reading a two-page pitch from Williams, he committed $50 million and later almost doubled his initial stake, according to WIRED. It wasn't his first bet on brain-adjacent technology. Bezos Expeditions previously backed Synchron, the brain-computer interface developer, in a $75 million round announced in December 2022.
GV confirmed its participation two months after the WIRED story broke, posting on LinkedIn: "We're honored to be part of Flourish's early funding and to support their mission."
The Competition (Such As It Is)
Few companies are attempting exactly what Flourish is attempting. Cortical Labs grows biological neurons on silicon chips. Merge Labs, which counts Sam Altman among its backers, describes its work as "bridging biological and artificial intelligence." Neither has disclosed funding anywhere near Flourish's scale.
The broader neuromorphic computing landscape includes Intel's Loihi chips and IBM's TrueNorth architecture, both designed around neural-inspired structures that consume far less power than traditional processors. But those efforts have remained largely in research mode, struggling to break into commercial applications where conventional GPUs dominate.
Whether brain-inspired AI can deliver on its efficiency promises remains an open question. Recht, the Berkeley professor advising Flourish, put it plainly to WIRED: "I'm not convinced that it's going to work. But if it does, it would be amazing."
That uncertainty hasn't deterred the investors writing nine-figure checks. Perhaps they're betting less on certainty and more on the possibility that Reardon has figured out something the rest of the industry missed while racing to build ever-larger models. Or perhaps they just like long shots with world-changing upside.
