Jeff Bezos has put money behind one of the most contrarian bets in artificial intelligence. Flourish, a neuroscience-focused AI startup, raised $500 million at a $2.5 billion valuation from a roster that includes Bezos, Lux Capital, GV (Alphabet’s venture arm), and Catalio Capital. The round closed around June 4. Bezos reportedly committed about $50 million before nearly doubling his stake as other high-profile investors piled in. What he is buying into is a direct challenge to how the entire AI industry currently works.
A Bet on Something Different
The dominant approach to AI is brute force: bigger models, more data, more chips, more power. Flourish is going the other way. It is building what it calls Cortex AI, a system designed to emulate brain function by mapping real neurons and their connections — a field known as connectomics — in a hunt for what the founders call the brain’s “core algorithm.” Instead of scaling artificial neural networks that only loosely resemble biology, The startup is putting actual neurons under the microscope to learn how the organ computes so efficiently, then trying to reproduce that in silicon.
Related: Dragon Season 3 Premieres Tonight Amid Naval Crisis
The team has the pedigree to be taken seriously.
Co-founder Thomas Reardon created Internet Explorer at Microsoft and later founded CTRL-labs, a brain-computer interface company Meta acquired in 2019 for an estimated $1 billion. Co-founder Rob Williams is a former Amazon senior executive. This is not a first-time team chasing a hype cycle.
Related: Breakthrough in Regenerative Medicine: Japan OKs Pioneering Therapy Using Repurposed Human Cells
The Power Gap That Drives the Pitch
Modern AI burns enormous energy because it runs vast amounts of matrix multiplication across thousands of power-hungry chips. Frontier models are trained and served in data centers drawing hundreds of megawatts, and that demand is now straining electrical grids and water supplies and driving up the cost of running models. The human brain, by contrast, performs reasoning, perception, and learning on roughly 20 watts — about what a dim light bulb uses.
That efficiency gap is the entire thesis. Neural tissue achieves it through architecture utterly unlike a GPU: sparse, event-driven signaling, massive parallelism, and computation co-located with memory, rather than shuttling data back and forth between separate processors and memory banks the way conventional chips do. Flourish’s wager is that if researchers can identify the principles behind that efficiency — the “core algorithm” — they can build AI that delivers comparable intelligence at a fraction of the energy. That would touch everything the current AI buildout is straining: electricity demand, data-center water use, and the per-query cost of running models.
Related: Xiaohongshu reportedly secures 2026 World Cup streaming rights
The Hard Part
This is early-stage, high-risk research, and it is worth being clear-eyed about that. Neuroscientists have spent decades trying to understand how the organ computes, and there is no guarantee a single “core algorithm” exists to be found, let alone copied into hardware on a useful timeline. It has raised money and assembled a strong team, but it has not yet shown a working system that matches its ambitions. Brain-inspired computing — including neuromorphic chips — has been pursued for years without displacing the GPU-driven approach. The funding buys runway for a long-horizon bet, not a product you will use soon.
Bezos and a set of top investors are wagering that the way past AI’s energy wall runs through neuroscience rather than more silicon. The appeal is obvious: an AI that thinks on a laptop’s worth of power would reshape the economics and the environmental footprint of the entire field. The risk is just as obvious: reverse-engineering the organ is one of science’s hardest open problems, and the startup has to solve it before any of the efficiency promise pays off. It is among the most ambitious AI bets of the year — and one of the least certain. For now, it is a thesis with serious backing, not a breakthrough.