What If the Chain Runs on Neurons? The Quiet Revolution Beneath Web3's Energy Crisis

There is a number that haunts every infrastructure conversation in this industry: 20 watts. That is the power consumption of the human brain. A single rack of servers in a modern data center can draw ten thousand. We have spent fifteen years building digital economies on silicon that screams for electricity, while the most sophisticated computing system we know hums along on the energy of a dim lightbulb. And now, a university in Singapore has decided to test whether we can build the next generation of infrastructure on biology itself. From the ashes of 2022, we planted seeds for 2030. But I did not expect those seeds to be neurons. The National University of Singapore recently announced what it calls the world's first data center powered by human brain cells. The headline is provocative, almost absurd. But beneath the clickbait lies a question that should matter deeply to anyone who cares about where this industry is heading: what happens when the substrate of computation stops being silicon and starts being something alive? Let me be clear about what this actually is, because the media framing obscures more than it reveals. This is not a data center in the traditional sense. It is a laboratory-scale experiment in biological computing, a field that sits at the intersection of neuroscience, stem cell biology, and computer engineering. The core technology involves induced pluripotent stem cells, or iPSCs, differentiated into brain organoids, tiny three-dimensional clusters of human neurons that can be cultured on electrode arrays. These organoids are not thinking in any human sense. They are computing in a primitive, distributed way, processing electrical signals through networks of living synapses that exhibit plasticity, the ability to strengthen or weaken connections based on activity. The field has a short but fascinating history. The most prominent pioneer is Cortical Labs, an Australian company that in 2022 unveiled DishBrain, a system of roughly 800,000 human brain cells cultured on a chip that learned to play the video game Pong. The cells were not programmed. They were trained, using feedback signals that rewarded successful predictions. It was a crude demonstration, but it proved something profound: biological neurons can perform goal-directed computation outside a body. Since then, a handful of players have emerged. FinalSpark in Switzerland offers remote access to organoid computing platforms. Koniku in the United States is engineering olfactory neurons for chemical detection. Stanford and Harvard have received DARPA funding for organoid intelligence research. The European Union poured over a billion euros into the Human Brain Project, though that effort focused more on simulation than on living tissue. NUS's contribution is not a breakthrough in the underlying science. It is a conceptual shift in application. By framing brain cell computing as data center infrastructure, they are asking a question no one has seriously asked before: can biological tissue become the backbone of the internet? This is where the story connects to us, to Web3, to the energy crisis that has shadowed this industry since its inception. We have spent years arguing about proof-of-work versus proof-of-stake, about energy consumption and carbon footprints, about whether decentralized networks can ever be sustainable. The debate has been framed entirely within the paradigm of silicon. We assumed the only path forward was more efficient chips, greener energy sources, better cooling systems. But what if the answer is not a better chip? What if the answer is a different kind of computation entirely? The energy math is staggering. A human brain performs roughly one exaflop of operations, a billion billion calculations per second, on 20 watts of power. The most efficient supercomputers on Earth require megawatts to approach similar throughput. The gap is not incremental. It is several orders of magnitude. If biological computing could be scaled, and that is a monumental if, the energy economics of every digital system we have built would be upended. Data centers would no longer be the bottleneck. The carbon cost of the internet would collapse. And the blockchain industry, so often criticized for its environmental footprint, would have a path to redemption that does not require compromising on decentralization. But I need to pause here, because my role in this community has never been to sell dreams. It is to ask hard questions. And the hard questions about biological computing are very hard indeed. The first is technical maturity. This technology is at Technology Readiness Level 3 or 4, which in plain language means it works in a lab under controlled conditions and has no credible path to production. Brain organoids typically survive for months, not years. They are noisy, their outputs are variable, and their error rates are far too high for the kind of deterministic computation that financial systems require. Scaling from thousands of neurons to the billions needed for meaningful data center workloads is not an engineering challenge. It is a biological one. We do not know how to keep that many cells alive, synchronized, and functional outside a body. We do not know how to read and write signals at the scale required. The gap between a laboratory demonstration and a production system is not five years. It is closer to fifteen, if it is ever crossed at all. The second question is ethical, and it is one I rarely see discussed in the coverage of this announcement. We are talking about using human neural tissue as infrastructure. The cells in these organoids are not conscious. They are not sentient. They are collections of neurons that exhibit electrical activity, nothing more. But the line is blurrier than we like to admit. Research has shown that organoids can develop spontaneous electrical patterns that resemble early brain activity. No one believes these structures are thinking. But as they grow more complex, as we engineer them to be more capable, we are moving toward a future where the distinction between computing tissue and living tissue becomes uncomfortable. Who consents to the use of their cells for this purpose? What happens when the organoids are discarded? Do we owe them anything? These are not questions that can be answered by a technical roadmap. They require a moral framework, and the industry has not built one. The third question is the one that keeps me up at night, and it is the reason I am writing this essay. We in Web3 have built our entire philosophy on the idea of decentralization, on the belief that power should be distributed, that no single entity should control the infrastructure of value. But biological computing introduces a new axis of centralization that we have not even begun to grapple with. Who owns the cell lines? Who controls the patents? The core intellectual property in this field is concentrated in a handful of institutions: Cortical Labs, Stanford, Harvard, and now NUS. If biological computing ever becomes the substrate of the internet, we will have traded one form of centralization, the hyperscale data center, for another, the biotech monopoly. The cells themselves are living things. They cannot be forked. They cannot be audited by a community of validators. They are proprietary in a way that code never is. I have spent the last year watching institutional capital flow into this space through Bitcoin ETFs and AI infrastructure funds. The narrative is always the same: efficiency, scale, progress. But efficiency for whom? Scale for what purpose? The promise of biological computing is that it could solve the energy problem that has haunted our industry. The danger is that it solves that problem by concentrating power in ways that make the current system look like a model of democratic governance. There is a contrarian view worth considering, and I want to give it its due. Perhaps the centralization risk is overstated. Perhaps biological computing will remain a niche research area, useful for drug screening and disease modeling, but never a serious competitor to silicon. The market for organoid intelligence in pharmaceutical research is real and growing. If that is the outcome, then the NUS announcement is not a harbinger of a new computing era. It is a public relations exercise, a university generating headlines to attract funding. The blockchain angle is incidental. The energy crisis remains unsolved, and we continue our slow march toward more efficient silicon, which is a fine outcome, just not a revolutionary one. But I cannot shake the feeling that we are at the beginning of something. Not because the technology is ready, it is not. But because the questions it raises are the questions we have been avoiding for a decade. What is the true cost of computation? What do we owe to the systems we build? And when we say decentralization, do we mean it as a technical property or as a moral commitment? I have audited enough protocols to know that most innovations fail. The ones that succeed are rarely the ones with the best technology. They are the ones that solve a problem people actually feel. The energy problem is real. The centralization problem is real. The ethical vacuum at the center of our industry is real. Biological computing does not solve any of these problems today. But it forces us to confront them, and that alone is worth something. From the ashes of 2022, we planted seeds for 2030. I did not expect those seeds to be neurons. But perhaps that is exactly what we need: a reminder that the most powerful computing system we know is not made of silicon, and that the future of infrastructure may not be built by engineers alone. It may be grown. And if it is grown, we need to decide, now, who gets to tend the garden. Trust is built in the bear, sold in the bull. But trust in what? In code, in consensus, in the invisible hand of the market. The next decade will test whether we can extend that trust to systems that are alive. I do not know the answer. I only know the question is coming, and it will not wait for us to be ready.