- AI model training consumes 6,000 times more energy than a European city, but bioprocessors promise significant energy savings.
- This article discusses how bioprocessors can revolutionize energy efficiency in computing.
In a strategic move, FinalSpark is poised to launch Neuroplatform, the world’s first living computer, in Switzerland. Instead of silicon chips, this groundbreaking bioprocessing platform uses human brain organoids, miniature replicas of lab-grown brains, to perform computing tasks. The company says the platform can house 16 brain organoids and uses only a fraction of the energy required by traditional silicon chips.
Silicon-based chips have revolutionized computing by enabling miniaturization and scalability but suffer from a lack of energy efficiency. Final Spark estimates that 10 GWh of energy was consumed to train GPT-3, the large language model used at the launch of ChatGPT. This amount is a whopping 6,000 times higher than the annual energy consumption of an average European city.
See more: Combatting AI Energy Consumption through Renewable Sources
Neuroplatform has addressed this problem by integrating human brain tissue into computational tasks and developing an innovative system called Wetware. This cutting-edge approach combines hardware, software, and biological elements through multi-electrode arrays (MEAs). In each MEA, four brain organoids are connected to eight electrodes, enabling simulation, data recording, and processing with a digital-analog converter.
The bioprocessor excels in energy efficiency over silicon chips yet encounters the issue of organoid mortality. Initially, the organoids had a lifespan of a few hours. However, FinalSpark’s advancements to the MEA systems have significantly extended the organoids’ operational life to 100 days.
FinalSpark makes Neuroplatform accessible to research institutions at a monthly rate of $500 per user. By collaborating with nine institutes, the company is advancing its mission to develop the world’s first fully operational living processor and revolutionize the landscape of biocomputing.