Swiss biotech startup FinalSpark has launched a unique online platform that, for the first time in history, provides remote access to a “living processor” – 16 organoids of the human brain. They act as biological processors that can learn and process information. Most such bioprocessors ““It consumes a million times less energy than traditional digital processors.”the company claims.

According to FinalSpark, neural platforms consume millions of times less energy than traditional electronic processors. For example, training a single Master’s language model such as GPT-3 requires approximately 10 GWh of energy; This is 6,000 times more than the average European consumes in a year in their daily lives. The use of bioprocessors will significantly reduce the huge energy costs associated with AI models and reduce the negative impact of computing on the environment.

The neuroplatform architecture is based on the Wetware concept, which combines hardware and software with biological components. Tom’s Hardware explains that it is based on four multi-electrode matrices (MEAs) into which living tissues (organoids, which are three-dimensional cellular masses of brain tissue) are placed.

Each matrix contains four organoids connected to eight electrodes for stimulation and signal recording. Data are transferred via Intan RHS 32 analog-to-digital converters at 30 kHz, and a microfluidic system and surveillance cameras are used to support the lifespan of the organoids. A software stack allows scientists to input data and read the responses from this unique bioprocessor.

Unlike silicon chips, which can last for years, the lifespan of a single living neuronal chip is approximately 100 days. Although the organoids initially lived for only a few hours, improving the life support system made it possible to significantly extend their active existence.

9 scientific institutions were granted remote access to the neuroplatform for biocomputer research. More than 30 universities are also interested in working with this revolutionary technology. For educational purposes, a subscription to the platform costs $500 per user. Commercialization of bioprocessors could usher in a new era of computing systems that are more ecological and closer to natural human intelligence.