As AI models grow larger, their energy consumption continues to soar projected to reach 3 percent of global electricity usage in five years. To tackle this challenge, researchers are turning to organoid intelligence, a new field that uses lab-grown brain cells to create energy-efficient computing systems. Scientists like David Gracias at Johns Hopkins University are developing biochips that combine living neural organoids with advanced hardware, aiming to mimic the brain’s dense 3D connectivity that far surpasses traditional silicon chips.
Gracias’s team built a 3D EEG shell that wraps around organoids, allowing more natural stimulation and recording of neural activity. Using reinforcement learning electrical pulses paired with dopamine rewards organoids can learn patterns and control devices such as miniature robot cars. This opens possibilities for robotics, prosthetics, disease modeling, and drug testing using organoids that mimic conditions like Parkinson’s disease.
However, major obstacles remain: biochips require constant care, suffer from signal noise, and depend on bulky lab systems. Companies like FinalSpark envision future bioservers built on living neurons, but scaling, programming neurons, and securing funding remain difficult. Still, researchers believe brain-like chips could eventually offer powerful, ultra-efficient alternatives to today’s silicon-based AI systems.