When Biology Becomes Compute, Energy Becomes Abundance

 The AI Energy Crunch: Why FinalSpark Thinks Neurons Are the Answers 

In this episode of The Disruptor Podcast, host John Kundtz sits down with Dr. Ewelina Kurtys, Strategic Advisor at FinalSpark, to discuss a hard truth most AI leaders are starting to feel: the next wave of AI scale will be constrained less by algorithms and more by energy, cooling, and infrastructure limitations.

While much of the industry is still running the traditional playbook, more GPUs, bigger clusters, and ever-larger data centers, FinalSpark is pursuing an entirely different path: biocomputing built from living neurons. 

Ewelina explains why biology may offer a credible route to radically lower-power computing, what FinalSpark is actually building today, and why the biggest breakthrough isn’t just hardware, it’s understanding how neurons encode information.

You’ll also hear how FinalSpark is opening the door for universities and private-sector R&D teams to experiment now through its remote Neuroplatform, why this moment feels like the early days of quantum computing, and what it will take to move biocomputing from lab credibility to commercial adoption.

Together, they explore what it means to stop brute-forcing scale and start rethinking compute from first principles, before the grid, the water, and the economics make the decision for us.

In this episode, you’ll learn:

1️⃣ Why the AI “energy crunch” is bigger than electricity—and increasingly about cooling and water

2️⃣ Why “just add more GPUs” is a short-term fix with long-term structural risk

3️⃣ What FinalSpark’s neuron-based biocomputing approach is—and what it can (and can’t) do today

4️⃣Why the hardest problem is decoding neural information, not simply measuring neural activity

5️⃣How R&D teams can start experimenting now via FinalSpark’s remotely accessible Neuroplatform

About our guest:

Dr. Ewelina Kurtys is a neuroscientist-turned-entrepreneur and Strategic Advisor at FinalSpark, working at the intersection of neuroscience, frontier computing, and commercialization. She has built a career translating complex science into market-facing strategy, bridging lab reality with business adoption.

Visit FinalSpark: finalspark.com
Connect w/ Ewelina on LinkedIn: 

Transcript

Introduction to the Disruptor Podcast

John Kundtz 0:00

The AI energy crunch. My final spark thinks neurons are the answers. Hi everyone, I'm your host, John Kundtz, and welcome to another edition of the Disruptor Podcast. For those that are new to our show, the Disruptor Series is your blueprint for groundbreaking innovation. We started the podcast in December of 2022. Our vision was to go beyond conventional wisdom by confronting the status quo and exposing the raw power of disruptive thinking. And today's guest embodies that spirit. Dr. Evelina Cortez is a neuroscientist, turned entrepreneur, and a strategy advisor at Final Spark, where they are building computers from living neurons. Today we'll explore how biocomputing works, why it matters, what neuron why neurons may be the most credible path to radically lower power compute. Welcome to the show, Dr. Kurtys.

Dr. Ewelina Kurtys 1:13

Thank you so much.

John Kundtz 1:14

As I said in our prep, this actually embodies the mantra or the tagline of the disruptor show, which is to really think uh beyond the status quo. And I think today we'll talk a lot about some of the things that particularly some of the large AI companies are doing that are probably fall into the status quo and some of the things that you are doing to really disrupt that space and hopefully actually make a big impact to the energy consumption and the grid and the future of both computing and particularly AI computing. But before we get started, why don't you just tell us a little bit about your background, your education, your experiences, how'd you get here? Please feel free to start anywhere you want.

Dr. Ewelina Kurtys 2:00

So I'm neuroscientist by background, as you mentioned. I'm from Poland. I have done research in different countries and PhD in the Netherlands. And later I end up in the UK, where I live now. And I moved from academia to industry because I was always curious about the world outside. I started to work with startups and I discovered the world of artificial intelligence. Become very fascinated by this. And now I work on AI only. So I left my research. I work with Final Spark as strategic advisor. This is a startup in Switzerland, which is building computers from living neurons. So this is a combination of neuroscience and engineering. So my two passion: one is from academic research, another from industry. So this is him where I am today.

The AI Energy Crunch and Its Challenges

John Kundtz 2:48

That's excellent. I actually followed a similar path 40 some odd years ago where I was a PhD student doing research in geochemistry and got into computers, and that led me to a lifelong career at IBM. I love your curiosity between research and business and enterprise as a very passionate for me, so it sounds like it's super passionate for you. But as you know, you can't turn on the TV, listen to anything, read anything that doesn't talk about AI today, which is one of the reasons I was super excited to have you on this show. And as we are according this, we are at a probably a pivotal growth where AI is exploding, and therefore the demand for power and cooling and data centers and just the constraints on the electrical grid is huge, particularly here in the United States. You see some of these frontier companies and many of the large tech companies, they're starting to build their own power plants. Some of them are recommissioning nuclear power. The demand is much larger than the supply. But more importantly, it's driving up the electrical rates for the average consumer. So I think some of the things you're trying to do very timely, and I'm sure our audience will be very interested in understanding more about it. But let's talk about first, sort of if you're a, I'll call it AI or a critical infrastructure leader, these folks seem to, as I mentioned, seem to be just defaulting to what I call the traditional playbook. Throw more GPUs at it, bigger clusters, try to incrementally improve the efficiency, build bigger data centers. And of course, this is all has this massive increase in the electricity consumption. So what do you see as the biggest challenges these these guys face as they start to deal with the AI energy crunch?

Biocomputing: The Future of AI Hardware

Dr. Ewelina Kurtys 4:42

Well, the biggest issue with the current AI is, I think, for environment is energy consumption and also water consumption. So it's not only about electricity. Actually, a huge data center. They also use a lot of water for cooling down. And these are important problems. And also so these problems are both for environment and also for the cost of the service. And today we don't feel it yet, maybe because a lot of AI services is underpriced. Like for example, ChatGPT, OpenAI is losing a lot of money. But it's a normal stage in every new technology that there is some time of adoption. So that means that prices are lower than they should be to make the business profitable. So we don't feel it yet, but AI is very expensive, and also the better the models, the more energy, actually, the more resources they need to run, to be trained and to run. So this is, I think, very important scalability problem because when we go forward, we want to use more and more AI and more advanced models, and this will become more and more expensive. So we want to solve this problem by changing the hardware. So getting out from silicon and changing silicon to biology, and we want to build totally new processors, bioprocessors, which will be built from living neurons. So instead of silicon, we will have living neurons.

John Kundtz 6:10

That's amazing. I'm still trying to get my head wrapped around the fact that we could use living neurons to replace silica. But you make some excellent points. I spent a good bit of my career in the data center strategy and design and build world at IBM. And one good point that you said, which I think people don't realize, is the electricity is only half of the equation. I always said for every dollar of electricity you spend, you're probably going to spend another dollar on cooling, which is typically water-based. And so those are those are interesting, you know, excellent points. So what I mean, what can people sort of do today? Is there any sort of mindset shift that you wish folks would do, leaders would do to adopt before doubling down on the same sort of compute and power playbook? Are there things they can do today based on what you're working on? And then we'll get into uh a little bit about how you're getting what you're doing in the lab, out of the lab and into the commercial world.

Dr. Ewelina Kurtys 7:07

I think the most important is to develop new technologies and we focus on that. We don't maybe want to change the way of thinking, but we want to provide some, build something new. And if it works, I think the arguments will be very strong to use it because if people will have access to AI, which will cost 10 times less, 10 times less, that I think will be an important argument. And I think it's a big challenge because if you don't use AI, you are behind those who use it. So I think we kind of, in my opinion, we don't have the choice but using AI. And I think it's not about restricting the usage of AI, but it's more I think we have to think about solutions on how to make it more sustainable. But I wouldn't encourage people to stop using AI because I think that's that's not a good idea.

Prototyping Bio Computers at FinalSpark

John Kundtz 7:55

As they sometimes say here, the US, so that that ship has left the port or that train has left the station. And you do make a good point earlier that I failed to reiterate, we're in the point now where much of the energy and the compute power is focused on the training models of the of the large language models, correct? And yes, that's like just the beginning. What we haven't experienced yet is the boom in the actual usage. So as people begin to use the models, there's a huge compute demand on that side of the equation. That that's going to be a logarithmic curve that we're at the bottom of it today. So I think people forget that this demand for power and energy is just scratching the surface. And and of course, the CapEx requirements, and I think you alluded to this with OpenAI, they're facing huge CapEx expenditures, which are really hard to make up when you're only charging $20 a month for a paid subscription or $200 a month for an enterprise license. Let's talk about what you guys are doing and what makes your approach different. And how do you plan to bridge between the stuff you're doing in the lab and getting it into what I call the boardroom where people can start to benefit or reap the benefits of using neuroscience as a compute resource?

Dr. Ewelina Kurtys 9:22

We are currently working on prototypes of our biocomputers. We aim to build huge bioservers in the future, which will be center computer, where you will be able to connect remotely the same way as today we connect to cloud computing. And of course, there can be many biocomputers in the future, many servers like that. And today we have very little prototypes, so that means we culture living neurons in the lab, and we have little round structures of living neurons, which consist of around 10,000 neurons each. And this is called organoids. They are such 3D structures of neurons. Some people call this brain in the dish, but I think it's misleading. It has it's not a brain, it's just the same building blocks, which are neurons, but it's far away from brain, because we would never be able to reproduce all the fine connections and structures which are in the human brain. But we can use the building blocks. So neurons we put on the electrodes, usually on eight electrodes. And you can see how it looks on our website, finalspark.com. There is section live. And we can, in this way, we can send them electrical signals, and we can also measure the response, which is also electrical signals from the neurons. And this is kind of like input and output, the same as in computers, and we want that this input and output make sense. So that we can process information via living neurons. That's our objective.

John Kundtz 10:48

I'm struggling with just it because this is really, I said this is just incredible. By the way, we will put a link to the website. I I spent some time on your website in the prep for the show. You can go deep dive into this stuff. It's fascinating. But if you think about it, neurons are living things that allow electrical electrons to move through the body, correct? That's how our nervous system works. It's it's probably not that much different from the physics standpoint as moving the electrons across silicon wafers. Am I thinking that am I correct on that? I'm not a biologist, I was a geologist.

Dr. Ewelina Kurtys 11:24

It's not only physics, but uh it's biology, which is much more complicated. Spikes, the electrical signals in neurons are different. They are not flows of electrons, but these are changes in ions. There are some minus ions on the membrane of the neuron, and they change the location, and that's why the voltage is changing. And this can be detected as a spike on the electrodes. So the mechanism is different. And actually, biological systems are usually very complicated. They are messy, they are unstable systems, dynamic, as we call in mathematics. And this creates a lot of problems because it's hard to do something reproducible, it's hard to control these systems. And there is a lot of unknown because nobody today knows how neurons encode information. So we know quite a lot how they process it, but we do not know how they encode. So basically, we can measure spikes, the activity electrical, but we do not know what it really means. And that's the area of active research, which is very challenging. But when we answer these questions, this will be revolution, not only for biocomputing, but also for medicine, for example, maybe computer brain interfaces. So I think this knowledge has a great potential once it's discovered.

Commercial Applications and Future Prospects

John Kundtz 12:38

It's fascinating. Is there anything that the commercial world should be doing, or is there any sort of bridging that we that they can do between now and when you break through through some of these technologies?

Dr. Ewelina Kurtys 12:51

Yes, absolutely. So, for example, our lab is available remotely, which is very special. We invited universities from all over the world to collaborate with us. And later we started to observe that the private sector gets interested. And we currently have an increasing number of users who pay us to get the access subscription. So it's called Neuroplatform, Final Spark Neuroplatform, and this is our lab on the cloud, kind of. You can connect physically to the physical lab in Switzerland, and you can do experiments the same way as our scientists are doing. Because today in Final Spark, you do experiments by writing code in Python programming language. So you can do this either in our lab physically or you can do this anywhere in the world. It's always the same. So people are interested to stay on the edge of science and they want to try this technology, even if it doesn't work yet. But if for today we can do very basic, I would say fundamental research on how neurons process, but people are already interested. So I think this is very cool for RD teams to stay ahead. And the same things actually happen in quantum computing field. This is a similar field in the way that you also have new, totally new type of hardware. And you know, it still doesn't work yet, but people believe in this. That's why many companies actually spend a lot of money to learn the prototypes so that they can be prepared once it will work for real.

Final Thoughts and How to Get Involved

John Kundtz 14:17

So I was actually just about to make that same analogy. It sounds very similar to what's going on in the quantum computing world. At IBM, we were heavily into quantum computing, and we did the same thing. We opened up our research labs to anybody that wanted to play an experiment. And I think you're absolutely right. These are baby steps. But at some point you're gonna make a breakthrough, much like Google did when they built the transformer. They had a lot of research, and then all of a sudden, boom, they created this transformer, which then opened up the whole world of generative AI. And I expect you're on the same path. At some point, you're gonna solve all these little problems to a point where it's gonna just go boom, and now you're ready to go. And I and what I see is this could be I mean, talk about disruptive. It's probably like going from vacuum tubes to transistors in in the computer world back in the whatever that was, the 40s and the 50s, when IBM was able to take their mainframes and put them onto transistors and then put those transistors into silicon wafers, allowed us to open up a whole new world of computing, which we, you know, nobody ever dreamed of. So this is fascinating. So there are definitely things people can be doing today. It sounds like if they want to play around and experiment and incorporate some of what you're doing into their research, they can get access to your labs and through the through the internet or through the cloud. Is there anything else you want to talk about or want to share?

Dr. Ewelina Kurtys 15:46

Yes, I definitely encourage people to check our website, finalspark.com. If you are interested in any aspects of our work, it's great to check our website. You can also get in touch. We answer every email we get. We also have Discord community for technical discussions. We have a newsletter for people who want to stay up to date. So we have a lot of resources to spread the message about our work.

John Kundtz 16:08

Excellent. I'll get those links and we'll put them into the show notes so anybody listening or reading any of our blogs will be able to get in contact with you. This has been such a rich conversation. We talked about how people can get in touch with you and your services. I know you're on LinkedIn. Would you like people to also reach out through your LinkedIn profile?

Dr. Ewelina Kurtys 16:30

Of course, absolutely.

John Kundtz 16:32

Good. All right. Excellent. Well, what I like to do is let's give you the last word before I wrap up the show, anything you'd like to say.

Dr. Ewelina Kurtys 16:41

So as mentioned, I encourage people to check our website. We build computers from living neurons, which will be one million times more energy efficient as compared to silicon hardware. And if you know want to know anything about the project, technical or about the future, commercial future, feel free to reach out to us and we are happy to answer any questions. And also we are currently fundraising, so maybe it's good to mention because the project is so far self-funded by the funders, but we are also talking to investors about 50 million Swiss francs investment. So we also have we are open to discuss with investors. Yes.

John Kundtz 17:16

That's a great point. Right now you're just funded by by the founders. Is it similar in the United States where you'd be raising an angel round or a seed round, or are you more like Series A? Or where are you with us?

Dr. Ewelina Kurtys 17:28

Well, we talk with anyone worldwide because we say 650 million Swiss francs. It's a lot of money. Very disruptive project because we expect to build computers in 10 years. So it's risky long term. But once it works, it will be very, very rewarding because we envision billions of profit, not only revenue, but profit actually. So this is not investment for everyone, of course, but we are open to discuss with anyone. We get actually a lot of risk requests on the website from people from US.

John Kundtz 18:02

Excellent. So there you go. I mean, the upside of this investment could be huge because the, as you mentioned, the impact is just in energy savings alone, could have not only great impact to our environment and to humanity in general, but can now bring along the the necessary compute resources to support the the butt the burst of demand in the AI world. This has been fantastic. It's so interesting. We could probably go on and on and get really detailed into the science end of it. But I want to thank you for joining. We will wrap this up. So again, I'm John Kundtz. Thanks for joining us in this edition of the Disruptor Podcast. Have a great day. Cheers.