The Little Brains Are Back

This is the second blog I’ve written about these little guys. The first was last fall, when brain organoids still felt like something sitting on the edge of science fiction. Tiny clusters of human brain cells in a dish, wired into computers, learning basic tasks. Interesting? Yes. Investable? Maybe someday.

Well, someday is starting to get closer.

Brain organoids are three-dimensional clusters of human brain cells grown from stem cells. They are not “brains” in the way we think of a brain. They are not writing poetry, worrying about interest rates, or asking whether the Fed is behind the curve. But they do form neurons, glial cells, and active neural networks. They produce electrical activity. They adapt. And, increasingly, researchers are giving them something that looks like a body — or at least a virtual one.

For years, artificial intelligence has mostly been about prediction. Feed a model enormous amounts of data, train it to recognize patterns, and then ask it to generate text, images, code, or answers. That has been impressive, and in some cases world-changing. But large language models do not have a body (at least not yet – see my series on the rise of robots). LLM’s do not have needs, they do not care if a signal is stable or chaotic, nor do they do not have homeostasis.

Brain organoids are different. They are biological systems. Neurons, like people, apparently do better when they have something useful to do.  Back in 2022, Cortical Labs made headlines when it connected living neurons to a simulated game of Pong. The system learned to move the paddle and return the ball through feedback. Naturally, that led to a wave of headlines about “sentience in a dish,” because nothing gets clicks quite like implying that a petri dish may be plotting against us.

That was overstated then, and it is still overstated now. These systems are not conscious in any meaningful human sense. They are not little people trapped in glass. But the field has moved beyond parlor trick territory. Researchers are now building closed-loop systems where organoids receive sensory-like input, act within virtual environments, and adjust their behavior based on feedback.

In one recent setup, cortical organoids were placed into a virtual “odor field.” The organoids effectively learned to move toward a food-reward zone, where they received more ordered and predictable electrical signals. In another, mouse cortical organoids were used in a Cartpole task, the classic balancing problem where the goal is to keep a virtual pole upright. The organoids showed goal-directed control and adaptation.

That does not mean they are “thinking” the way we think. It does mean they are learning in a fundamentally different way than silicon-based AI.  The human brain runs on roughly 20 watts. Data centers, as anyone following AI infrastructure knows, do not. We are building out enormous compute capacity, power generation, cooling systems, transmission lines, chips, memory, networking equipment, and software layers to support the AI boom. That is real, and I continue to think it is one of the defining investment themes of our time.

But biology has been doing low-power, adaptive computation for a very long time. We are the new kids on the block. Nature has been in the business for a few billion years.  That is why the idea of biohybrid intelligence is so interesting. It is not that brain organoids are about to replace Nvidia chips, hyperscale data centers, or large language models.  But in certain narrow applications — especially adaptive, embodied, data-sparse environments — biological systems may eventually offer advantages that traditional AI struggles to replicate efficiently.

One of the more interesting developments is Brainoware. In this approach, researchers place a brain organoid on a multielectrode array and use it as a kind of living reservoir computer. The organoid processes information through its natural neural dynamics, while a conventional readout layer interprets the result.  That sounds like something a bond trader would pretend to understand at a conference cocktail party. But the basic idea is straightforward: let the biology do what biology does well, then use traditional computing to read the output.

Brainoware has already been used for tasks like speech recognition and nonlinear equation prediction. Just as important, when researchers blocked synaptic plasticity, the system stopped improving. In plain English, the living part mattered. The organoid was not just decoration. It was part of the computational engine.

This is where the broader concept of Organoid Intelligence comes in. The vision is to build scalable biological computing platforms that can complement today’s AI systems. That is a big vision, and we should be careful not to get carried away. This field still has major obstacles: scalability, stability, vascularization, standardization, ethics, and cost. There is a long road between “interesting lab result” and “commercial platform.”  But long roads are where some of the best investment themes begin.

The more immediate applications may be in biotech and medicine. Better brain organoids could improve disease modeling, drug testing, and personalized medicine. Instead of relying as heavily on animal models, researchers may be able to test therapies on human-like neural tissue. That could matter in areas like Alzheimer’s, Parkinson’s, autism research, psychiatric drugs, and neurodegenerative disease.

If these platforms become more predictive, they could make drug development faster, cheaper, and more targeted. Anyone who has followed biotech knows how badly that is needed. Drug discovery is expensive, failure rates are high, and the human brain remains one of the hardest systems in medicine to model.

There is also a longevity angle, a personalized medicine angle, and eventually a computing angle. That is a pretty interesting neighborhood for capital to be watching.

Of course, this also comes with ethical questions. If we keep making organoids more complex, more connected, and more embodied, where are the boundaries? At what point do we need new standards for consent, moral status, or research limits? Today’s organoids are still far from anything resembling human consciousness. But the point of ethics is not to show up after the horse has left the barn, hired a lawyer, and started a podcast.  The public will have questions. Regulators will have questions. Investors should have questions too.

That does not make the field uninvestable. It makes it a frontier and frontiers are messy. They are usually overhyped early, dismissed by skeptics, misunderstood by the public, and underestimated by incumbents. Then, if the technology is real, it quietly improves until one day it seems obvious in hindsight.

That is how a lot of major investment themes develop. AI looked like a science project for decades. Gene sequencing looked expensive and niche. Electric vehicles were a punchline until they were not. Space was a government program until private companies rewrote the economics.  Now we have tiny brain organoids learning inside virtual worlds, responding to feedback, and hinting at a future where the line between biology and computing gets blurrier.

Again, these little guys are not managing your portfolio. They are not pondering the meaning of life. They are not asking whether small caps are finally cheap enough to matter. But they are another reminder that innovation rarely arrives in a straight line. Sometimes it comes through software. Sometimes it comes through chips. Sometimes it comes through biology sitting in a dish, learning in a loop.

Biohybrid intelligence sits at the intersection of several powerful themes: artificial intelligence, biotechnology, personalized medicine, longevity, neuroscience, and next-generation computing. That does not mean we should chase every company or every pitch deck with the word “organoid” in it. Quite the opposite. Early frontiers require discipline, skepticism, and a strong stomach.  But they also require attention.

The future of computing may not be purely silicon. It may not be purely biological either. More likely, it will be some strange hybrid that looks obvious only after someone else has already made the money.

As always, if you would like to talk about how this may affect your portfolio, or how we are thinking about AI, infrastructure, software, and the companies powering this next wave, please give us a call. Your capital, our expertise, a bespoke creation.