In Melbourne, entire racks of devices are already up and running, linking cultures of human neurons with silicon circuits.<\/p>
In October 2022, Brett Kagan’s team at Cortical Labs published in Neuron findings that read like a rejected science-fiction script. About 800,000 neurons\u2014some cultures derived from human induced pluripotent stem cells (sounds pretty good, doesn’t it?), others from primary cortical neurons of mouse embryos\u2014were placed on a microelectrode array. The researchers wired this setup into a simplified version of the game Pong. The system received information about the ball via electrical pulses. The paddle’s movements depended on the neurons’ activity. The authors observed signs of learning within five minutes of play. That’s not proof that the culture understood the rules of table tennis. The layer of living cells could simply change its activity as a result of coupling with the environment.<\/p>
The article\u2019s title mentioned feeling\u2014or even a kind of mini-consciousness. Many neurobiologists thought that wording was too bold. In March 2023, a group of researchers led by Fuat Balci responded in the pages of the same journal. They argued that Kagan\u2019s results don\u2019t justify inferring intelligence or sentience in the usual sense of those words. It\u2019s safer to talk about plasticity and learning in a closed stimulus loop. That\u2019s a big enough achievement. We don\u2019t need to tack consciousness onto it.<\/p>
Three years later, the company unveiled the CL1, a biocomputer that integrates neurons with a life-support system. The device debuted on March 2, 2025, at Mobile World Congress in Barcelona. One unit cost about $35,000, but buying them in racks of 30 brought the price down to about $20,000 per module. In March 2026, the company announced the launch of a prototype data center in Melbourne equipped with 120 such units. Each contained about 200,000 nerve cells\u2014though company materials from 2025 still cited 800,000. That discrepancy suggests the exact figures are somewhat fluid. It’s safer to speak in terms of hundreds of thousands of neurons.<\/p>
We have to be precise about power consumption. The NVIDIA H100 in its SXM version draws up to 700 W. The PCIe version has a configurable TDP in the 300-350 W range. According to the company, one CL1 unit draws about 850-1000 W. In March 2026, Hon Weng Chong stated that the actual compute module in the data center draws about 30 W. Those numbers look attractive, but we shouldn’t treat them as a simple performance ranking. CL1 won\u2019t replace an AI accelerator, as it\u2019s used to study living tissue and pharmacological responses.<\/p>
In March 2026, the Bio Data Centre in Melbourne launched. A larger center in Singapore is also planned in partnership with the operator DayOne. The first phase includes a single rack with 20 units at the Yong Loo Lin School of Medicine (NUS Medicine), ultimately scaling to a thousand units at DayOne’s commercial facility. This isn’t a biological alternative to the data centers that run AI chat services, but a demonstration of scale. It’s about making cell cultures available via the cloud, which lets researchers work without their own cell-culture lab. Remember The Matrix? <\/em>Harnessing brains for compute centers would be far more sensible than using human bodies as batteries.<\/p> A competitor operates in Switzerland. On May 15, 2024, FinalSpark announced the launch of the Neuroplatform, a system that offers remote access to experiments on human brain organoids. The platform lets you interface with 16 organoids arranged on four electrode arrays. You can control the system using Python and Jupyter notebooks. Claims of million-fold lower energy consumption mostly come from marketing. The science points more to a hypothesis about the high efficiency of biological networks. No real advantage over silicon has been demonstrated yet in complex computations.<\/p> After the initial excitement, it’s time for realism: these aren’t brains. FinalSpark organoids are about half a millimeter in diameter and contain around 10,000 neurons each. They lack the vascularization typical of in vivo tissue. They don’t have bodies, senses, needs or a history of interactions with their environment. The current models are simplified. They suffer from cellular immaturity and a lack of clear laminar structure. They’re tools for toxicity and drug testing, not candidates for conversation partners.<\/p> But even today we’re left with obvious questions about the future. What will we do if these constructs ever start to feel anything? Andrea Lavazza has called human brain organoids a double-edged sword. They provide better disease models than mice, but their growing complexity forces a reckoning with their moral status. Bioethicists warn against waiting to have this conversation until technology surprises us. We don’t have a simple test for consciousness. Tononi’s Integrated Information Theory focuses on the integration of information within a system. Dehaene’s Global Workspace Hypothesis links consciousness to the wide broadcast of information across a network of neurons. Graziano’s Attention Schema Theory treats the subjective state as an internal model of one’s own attention. Each framework points in a different direction. but none offers an easy playbook for assessing a brain organoid’s moral interests.<\/p> Science fiction writers have long explored these boundaries. Peter Watts, in Blindsight<\/em>, shows that high intelligence can exist without an inner sense of self. He also aptly portrays the management of the brain\u2019s computational power in exchange for a virtual paradise. Greg Egan, in Permutation City<\/em>, depicts digital copies of minds fighting for the status of their own existence. Debates over the moral status of biocomputers will oscillate between these visions. But the practical constraints are hard limits. A biocomputer won’t replace servers running language models. Living tissue needs nutrients, sterility and temperature control. It has a limited lifespan. Neurons in Cortical Labs systems survive up to half a year. FinalSpark reports a little over a hundred days. But it’s still an excellent environment for testing neurobiology, though not necessarily the best foundation for a biological ChatGPT.<\/p> There’s an important ethical argument here. Human neurons in vitro may model human diseases better than animals can, but that doesn’t eliminate the moral problems. It just changes their direction. Instead of asking about the suffering of mice, we’ll start thinking about protecting tissues that respond and learn in an artificial environment. Throwing away an old laptop is an inconsequential act. Shutting down a culture grown from human stem cells is still similarly simple. I just wonder how long the word “still” will apply here.<\/p>","protected":false},"excerpt":{"rendered":" When I tossed out my old laptop a few years ago, I didn’t give a thought to its possible suffering. Silicon is silicon. It cooled down and ended up in e-waste, case closed. Now that simple instinct is becoming problematic.<\/p>\n","protected":false},"author":247,"featured_media":19757,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[796,805],"tags":[],"popular":[],"difficulty-level":[37],"ppma_author":[614],"class_list":["post-20009","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-hai-premium-2","category-law-and-ethics","difficulty-level-hard"],"acf":[],"authors":[{"term_id":614,"user_id":247,"is_guest":0,"slug":"prof-dr-hab-dariusz-jemielniak","display_name":"prof. dr hab. Dariusz Jemielniak","avatar_url":{"url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/maxresdefault-1-e1742292469999.jpg","url2x":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/maxresdefault-1-e1742292469999.jpg"},"first_name":"Dariusz","last_name":"Jemielniak","user_url":"","job_title":"","description":"Profesor zarz\u0105dzania Akademii Leona Ko\u017ami\u0144skiego, gdzie kieruje katedr\u0105 MINDS (Management in Networked and Digital Societies). Pracuje te\u017c jako faculty associate w Berkman-Klein Center for Internet and Society na Harvardzie. Wiceprezes Polskiej Akademii Nauk. Cz\u0142onek Rady Programowej CampusAI."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/20009","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/247"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=20009"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/20009\/revisions"}],"predecessor-version":[{"id":20010,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/20009\/revisions\/20010"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/19757"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=20009"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=20009"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=20009"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=20009"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=20009"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=20009"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}