Startup creates bioprocessor platform using neurons

Swiss biocomputing start-up FinalSpark has launched the first online platform that will give researchers anywhere in the world the ability to carry out experiments remotely, 24 hours a day, using in vitro biological neurons as bioprocessors.

Neuroplatform will offer 24-hour access to 16 human brain organoids. These bioprocessors, made up of living neurons capable of learning and processing information, consume a million times less energy than traditional digital processors, potentially reducing the impacts associated with the increasing use of computing power, as is the case with Generative AI models, according to the startup.

FinalSpark says that three dozen universities have expressed interest in using the platform and has already granted free access to nine institutions for research use only. “As demand for our neuroplatform grows, we will be able to scale up the platform with the common goal of building the world’s first living processor. We firmly believe that such an ambitious goal can only be achieved through international collaboration,” says Dr Fred Jordan, co-founder of FinalSpark.

For FinalSpark, the convergence of Artificial Intelligence, recent advances in biology and stem cell technologies has opened up new horizons in the field of synthetic biology and so-called wetware, which uses computers composed of organic materials. “With the launch of our neuroplatform, we are at the forefront of this exciting journey. It’s an inspiring moment for us researchers,” emphasises Martin Kutter, co-founder of FinalSpark.

A scientific publication detailing FinalSpark’s neuroplatform is available in the journal Frontiers.

How it works

FinalSpark’s Neuroplatform is based on a wetware architecture, a mix of hardware, software and biology that uses four Multi-Electrode Arrays (MEAs) that house living tissue, i.e. organoids made up of brain tissue cells. Brain organoids are small clusters of neurons grown from stem cells.

In each MEA, four organoids are connected to eight mini-electrodes. These small electrodes capture measurements of cellular activity in real time, making it possible to simulate, record and process data using an analogue-to-digital converter. The platform also includes Python-based software so that users can interact with the bioprocessors.

In short, this bioprocessing platform uses 16 of these organoids to carry out computational tasks, with the advantage of consuming only a fraction of the energy required by traditional silicon chips.

Because they use living tissue, these bioprocessors face the problem of organoid mortality, which initially had a lifespan of a few hours. However, FinalSpark claims to have made advances in MEA systems and managed to extend this operational shelf life to 100 days, according to the SpiceWorks website.

Environmental contribution

On the company’s blog, FinalSpark says it believes that unconventional computing is the best way to contribute to environmental issues by reducing CO2 emissions. It emphasises that one of the main advantages of biological computing provided by Neuroplatform is precisely the fact that neurons process information using much less energy.

FinalSpark estimates that neuron-based bioprocessors could use a million times less energy than their silicon counterparts. “When we compare them with the fastest computers today, such as the Hewlett Packard Enterprise Frontier, we can easily see that, at approximately the same speed and with a thousand times more memory, the human brain uses only between 10 W and 20 W, while the computer uses 21 MW,” says the post.

Of course, using living neurons to build bioprocessors is not an easy task. What’s more, we still don’t know how to programme them. A lot of experimentation is still needed, and this is precisely what FinalSpark Neuroplatform aims to facilitate – collaboration between researchers to unravel and control this black box full of powerful computational tools.