The hyperscalers require thousands of chips and gigawatts of energy, enough to power small cities, according to a report in WIRED. That is the scale of the energy problem facing today's large language models (LLMs), which have become increasingly powerful but also voracious consumers of compute power and data. Now a new startup, Flourish, backed by Jeff Bezos, is attempting to build a fundamentally different kind of AI that mimics the efficiency of the human brain.
The Energy Crisis in AI
Today's frontier models have little in common with the human brain, WIRED reported. A person uses about 20 watts of energy to process information; a single chip in an AI training cluster uses more than 30 times that amount — over 600 watts. And those models need to absorb virtually all of human-written text. Each new model requires more data and more compute. Once trained, they do not learn continuously.
Thomas Reardon, Flourish's co-founder and a neuroscientist, told WIRED: "There’s something fundamentally wrong with saying, ‘I need to basically read every book ever written 20 times over in order to learn English.’ A human baby does it with a couple hundred thousand utterances." The goal is to build "a synthetic artificial intelligence brain that runs on 50 watts or less" that adapts to its conditions and uses a tiny fraction of an LLM's compute power.
The contrast in power consumption is stark:
| Component | Power Consumption | Notes |
|---|---|---|
| Human brain | ~20 watts | Biology's baseline |
| Single AI training chip | >600 watts | 30x human brain |
| Hyperscaler cluster | Gigawatts | Enough for small cities |
The Flourish Approach
Flourish describes itself as a "neuro AI company" that is solving what it calls the two most difficult problems facing AI today: power efficiency and continuous learning. The company is building Cortex AI, a synthetic intelligence system designed to match the computational capacity, learning efficiency, and power budget of the human brain. According to WIRED, Reardon and his co-founder Rob Williams have not yet figured out how to build such a system, but they believe an expert team of AI researchers and neuroscientists working side by side can find the answer. The neuroscientists will conduct original wet lab experiments with advanced lab equipment to hunt for usable intelligence on the brain's architecture. The company plans to release near-term models as products on the path to a full reinvention of AI.
The Team and Backing
Rob Williams, a former executive on Amazon's "S-team" who oversaw software products including Alexa, left Amazon in fall 2025. In December 2025, he pitched Jeff Bezos on the Flourish idea. According to WIRED, Williams knew how to pitch Bezos: you write a press release as if your product has already been built, and Bezos gives a thumbs up or down. After reading Williams's two-pager, Bezos chipped in $50 million. Other funding came from Lux Capital, Google Ventures, and Catalio. Bezos then almost doubled his initial stake and told Reardon he would have given more if they had asked. Flourish now has a war chest of $500 million and a reported valuation of $2.5 billion.
Thomas Reardon IV — he goes by Reardon — has a colorful background. He dropped out of the University of New Hampshire at age 15, became a teenage programming wizard, helped build Microsoft's first web browser, and started and sold a wireless tech company. He then earned a degree in classics from Columbia University, followed by a doctorate in neuroscience. Later, he co-founded a company that developed a mind-control wristband, which was acquired by Meta, where he worked for six years.
The Path Forward
Flourish's approach is a high-risk, high-reward bet on a biological-inspired alternative to the current AI paradigm. If successful, the company could deliver AI systems that are orders of magnitude more energy-efficient and capable of continuous learning — a potential game-changer for enterprise applications where power and data costs are escalating. For CTOs and technology decision-makers, Flourish's progress could signal a shift away from the ever-larger LLMs toward more sustainable, brain-like architectures. However, the company has a long road ahead, and its near-term products will be closely watched by the industry.
Sources:
