Wetware Computing Paves the Way for Low-Power AI

As artificial intelligence (AI) continues to advance, the energy demands and computational load on traditional silicon-based chips are skyrocketing. With conventional hardware approaching its physical limits, scientists are exploring novel approaches that integrate biology with technology. One promising direction is Wetware Computing, which leverages living neuronal networks to perform computations with remarkable energy efficiency and learning capability.

A recent review published in Advanced Science by research team led by Prof. CAI Xinxia and Assoc. Prof. LUO Jinping from the Aerospace Information Research Institute (AIR) of the Chinese Academy of Sciences highlights In Vitro Brain-Computer Interface (BCI) technologies as a key avenue in Wetware Computing. The paper emphasizes how these brain-on-a-chip platforms—supported by projects such as the National Natural Science Foundation Innovation Group Research—could redefine the way AI systems are built, offering low-power, adaptive, and sustainable alternatives to traditional computing.

At the core of this emerging field is In Vitro BCI cultured with neural cells network or brain organoids. These organoids, miniature brain-like structures derived from stem cells, can be cultured in two- or three-dimensional configurations, with three-dimensional networks providing higher computational and storage capabilities due to their complex structure. Paired with high-density microelectrode arrays (MEAs) that record and stimulate electrical activity, and microfluidic platforms that supply nutrients and remove waste while simulating physiological conditions, these systems form the foundation of brain-on-a-chip technology.

"Brain-on-a-chip platforms allow us to harness the incredible learning and adaptability of living neurons," said Prof. Cai. "This could lead to AI systems that are not only more efficient, but also capable of performing tasks in ways conventional hardware cannot"

Early studies have demonstrated that these platforms can be trained for tasks such as speech recognition, robotic arm control, and pathfinding. The organoids' ability to adapt and learn highlights the potential of Wetware Computing to complement or even surpass traditional silicon-based approaches.

Despite these promising developments, challenges remain. Scientists need to create more complex organoids, improve the scalability of microfluidic systems, and refine electrical recording and stimulation techniques. Prof. Luo emphasizes, "Advancing Wetware Computing will require interdisciplinary collaboration across neuroscience, engineering, and computer science, but the potential payoff in energy-efficient, intelligent AI is enormous."

The review concludes that In Vitro BCI and brain-on-a-chip systems represent a transformative direction in computing. By combining biology with technology, these platforms could shape the future of low-power AI, opening the door to adaptive, efficient, and sustainable computing solutions.