In labs around the world, tiny spheres of neural tissue are learning to recognize patterns and respond to electrical pulses—no biological brain required. Biotech companies like Cortical Labs and FinalSpark are growing brain organoids on multi-electrode arrays and connecting them to hardware shells, creating a new category of computers that operate on the same biological principles as human thought but with a fraction of the energy cost. The work represents a fundamental shift in how scientists think about computation itself, treating living neural tissue as a functional computing substrate rather than something to study in isolation.
This matters because artificial intelligence has become a voracious consumer of electricity. As AI systems grow more powerful, their energy demands have grown proportionally, with large language models requiring enormous cooling systems and electrical infrastructure. Biocomputers, by contrast, harness the brain's natural efficiency—the human brain uses roughly 20 watts of power to perform tasks that would require megawatts of conventional computing hardware. According to Brett Kagan, Ph.D., Cortical Labs' chief scientific officer, these systems can learn with far less data and more chaotic data compared to artificial intelligence, a capability that mirrors how human brains actually work.
The practical applications are already emerging across multiple fields. FinalSpark and Cortical Labs have adopted a cloud-based model, allowing researchers around the world to access biocomputing hardware remotely and run experiments without building their own facilities. Drug discovery researchers have begun using these platforms to test how experimental medications affect neural learning, collapsing what typically takes months into compressed timelines. Some biocomputers have even been trained to play video games, demonstrating their capacity for adaptive learning in complex environments. Thomas Hartung, a Johns Hopkins professor and expert in the field, believes biocomputing could serve as a stepping stone toward neuromorphic engineering—artificial neurons designed to mimic the structure and function of human brains.
But this technology does not arrive without complications. Brain organoids raise the same bioethical questions that have long shadowed stem cell and organoid research: What is the moral status of increasingly sophisticated neural tissue? At what point might these systems develop consciousness? Who owns the genetic information contained in organoids, and what rights do tissue donors retain? Scientists have taken a proactive approach to these concerns, consulting with bioethicists before widespread adoption rather than after problems arise. The concerns are real and worth taking seriously, but they have not halted research.
The current frontier is unpredictability. Organoid activity remains variable and difficult to train, complicating efforts to make biocomputers reliable for industrial use. As researchers deepen their understanding of how neural organoids behave and respond to stimulation, these technical obstacles should gradually dissolve. A study published in the Journal of Medical Internet Research synthesizes the current state of the field, separating genuine promise from hype. What emerges is a technology that is genuinely nascent but genuinely promising—one that could reshape biomedical research and computing itself, if scientists can solve the engineering challenges that currently stand in the way.