Final Spark’s Biocomputers: How a Swiss Startup Is Building the World’s First Living Computers

Artificial intelligence is hungry. Training today’s AI systems requires massive amounts of energy, often on the scale of megawatts. However, in Switzerland, a startup called Final Spark is exploring a radically different approach: using living neurons to build computers that are a million times more energy-efficient than silicon.

It sounds like science fiction, but it’s already here.

Final Spark and the Promise of Biocomputing

Final Spark is pioneering the field of biocomputing, also known as organoid intelligence. Instead of relying on silicon transistors, their approach utilizes brain organoids: tiny, lab-grown clusters of neurons that can fire, adapt, and form connections much like the human brain. These organoids are not just biological curiosities; they behave as living computers, capable of processing information in ways that traditional chips cannot replicate.

What makes this significant is the timing. As AI energy consumption continues to soar, researchers are hitting a wall with silicon-based systems. Final Spark positions biological processors as a potential solution: computers that are far more energy efficient, inherently adaptive, and closer to how real intelligence functions.

Inside Final Spark’s Neuroplatform of Living Computers

The Neuroplatform connects 16 organoids through multi-electrode arrays (MEAs). Each organoid is paired with eight electrodes that stimulate neurons and record their responses.

These organoids act as biological processors. Unlike silicon chips, which follow fixed instructions, they adapt, reorganize, and evolve much like a living brain.

This blending of biology and technology gives Final Spark’s processors a unique advantage: they are both computationally powerful and remarkably energy efficient.

Remote Access to Biological Processors

What makes Final Spark unique isn’t just their lab work. They’ve built a remote biocomputing platform that researchers can access 24/7.

Through APIs and Jupyter notebooks, scientists worldwide can:

• Stimulate neurons with electrical signals
• Record neural activity in real time
• Adjust variables like pumps or UV light for neurotransmitter release
• Run long-form experiments without setting foot in Switzerland

So far, Final Spark has maintained organoid viability for more than 100 days and collected 18 terabytes of data across over 1,000 experiments.

How Final Spark Enables Global Biocomputing Research

By opening the Neuroplatform to universities and research institutions, even offering free access to select academic labs, Final Spark has positioned itself as a hub for collaborative discovery.

Instead of a closed system, their vision is a living computer in the cloud, one that researchers across the world can log into, experiment with, and learn from.

Why Living Computers Could Solve AI Energy Consumption

The energy cost of AI is skyrocketing. Training today’s largest models consumes electricity on par with entire towns.

Final Spark claims that their biological processors use a million times less energy than silicon. That means a future where AI can expand without overwhelming the world’s energy grid.

The human brain is proof: it runs on just 20 watts, which is less than a light bulb.

The Case for Sustainable Computing With Organoid Intelligence

Biocomputing isn’t just about efficiency. It also holds promise for:

• Adaptive learning: Organoids can reorganize, making them more flexible than static chips.
• Medical breakthroughs: Because they’re built from human neurons, organoids can model diseases, memory, and learning in ways simulations can’t.
• A new paradigm: Just as silicon replaced vacuum tubes, organoid intelligence could replace or augment silicon as the substrate of computation.

This is where sustainable computing and radical innovation meet.

The Limits of Biological Processors Today

Of course, the technology is not without hurdles. For all of its promise, biocomputing is still in the early experimental stage, and there are significant limitations that must be overcome:

• Scale: Human brains contain about 86 billion neurons. Current organoids max out at thousands to millions, impressive for a lab dish, but still orders of magnitude away from matching brain-level complexity. Scaling them up without losing stability or function is one of the hardest scientific challenges.
• Variability: Unlike silicon chips, which can be manufactured with near-perfect consistency, living systems are unpredictable. Each organoid develops slightly differently, making it hard to standardize results or replicate experiments at scale.
• Fragility: Organoids require precise conditions to survive — nutrients, temperature, oxygen — and even with advanced perfusion systems, they eventually degrade. For long-term sustainable computing, extending organoid lifespans will be critical.
• Ethics: As organoids become larger and more sophisticated, they raise difficult questions. Could they one day exhibit a rudimentary form of awareness? And if so, what rights would a “living computer” deserve? These debates aren’t theoretical; they will grow louder as organoid intelligence research progresses.

These challenges make biocomputing a long-term bet rather than an immediate replacement for silicon.

Ethical Questions Around Organoid Intelligence

The use of living human neurons in computation is bound to spark debate. If organoids adapt and respond in increasingly complex ways, could they eventually achieve some level of awareness?

For Final Spark, the ethical frontier is as important as the technical one. The company frames its work as exploratory — a way to push the boundaries of neuroscience and computing while inviting global collaboration and scrutiny.

Final Spark’s Vision for Biocomputing and Sustainable AI

Co-founder Dr. Fred Jordan puts it bluntly: “We are running into a wall with the power consumption from new AI systems. It’s not sustainable at all.”

This candid observation captures the driving force behind Final Spark’s mission. According to Jordan, efforts to simulate human-level neural networks digitally — billions of neurons firing with thousands of connections each — would require the output of a small nuclear plant. In stark contrast, our biological neurons use just about 20 watts, making them potentially 1 million times more energy efficient than current silicon-based AI solutions. It’s this gap that inspired the pivot from software simulations to real human brain organoids as the substrate for computing. By doing so, they hope to pioneer a computing model that is efficient, adaptive, and sustainable.

Why Living Computers Could Redefine the Future of Computing

Final Spark’s research may still be in its infancy, but its implications are massive. The company is proving that biological processors are not just a scientific curiosity; they could reshape the foundations of computing itself. If successful, this work could:

• Transform AI energy consumption by showing that living neurons can compute at a fraction of the cost of silicon, making large-scale AI development sustainable.
• Redefine the meaning of sustainable computing by offering a model that doesn’t require megawatts of power and endless data centers, but instead mirrors the unmatched efficiency of the human brain.
• Open entirely new frontiers in neuroscience and technology by giving researchers real-time access to living brain tissue, potentially accelerating discoveries in both computing and medicine.

The takeaway is clear: Final Spark isn’t just building a new processor. They are questioning the very foundation of what computing is, and pushing us to consider what it could become in a future where living computers are as common as silicon chips.

This post was first featured on my Medium blog here – https://medium.com/p/97ac3b07dd39.