In 2018, the Vancouver-based start-up Nectome made headlines when Sam Altman paid a $10,000 deposit to join the waiting list for a service that it still does not provide: he wants to “freeze†his brain. Specifically, he wants to preserve his brain minutes after his death so that it can eventually be digitised to render him immortal. Altman, now CEO of the AI company behind ChatGPT, became the 25th person to join the waiting list. “I assume my brain will be uploaded to the cloud,†he told MIT Technology Review.
The idea of preserving a person's brain to revive it at some point in the future sounds like science fiction. Preserving a human brain in a such a way that it retains its microscopic structure poses major technological challenges. Subsequent stages of the process – should we ever reach them – pose even greater conundrums, and raise philosophical questions about the nature of consciousness.
Yet, according to a recent survey conducted by Ariel Zeleznikow-Johnston, a neuroscientist at Monash University in Melbourne, Australia, and colleagues, more than one quarter of physicians in the United States “found preservation somewhat or very plausible for enabling future revivalâ€. Asked to estimate the probability that, under the right conditions, the process “could retain sufficient neural information for future revivalâ€, the median response from the more than 300 doctors surveyed was 25.5 per cent.
Research studies published earlier this year describe what some in the field are calling major advances in brain cryopreservation technology, and Nectome is now one of about half a dozen companies offering brain or whole body cryopreservation services. Should we be taking this industry seriously, along with its ethical implications?
*
The quest for immortality is as old as time itself, but cryonics – the practice of freezing dead bodies or brains – was popularised by a 1967 book entitled The Prospect of Immortality. Its author, the physicist Robert Ettinger, founded the Cryonics Institute in 1976, which pioneered the process and continues to offer cryopreservation for $28,000-$45,000. The company now has more than 250 “patients†in cryostasis at its facility in Clinton Township, Michigan.
The cryonics procedure involves removing blood from the patient's body and replacing it with a solution of cryoprotectant chemicals. The body is then cooled down to -196°C with liquid nitrogen. The cryoprotectants act like biological antifreeze, preventing the formation of ice crystals during cooling. Instead, this causes the water in the body to solidify into a glass-like state, in a process called vitrification.
The Cryonics Institute claims that its “patients†can remain in cryostasis indefinitely, on the premise that “doctors of the future … have perfected techniques for repairing and restoring the human body… [and can] successfully revive persons who were considered deceased by today's standardsâ€.
But perhaps we don't have to learn how to revive our physical brains, in order to give them a “second lifeâ€. Brain cryopreservation researchers today tend to have a different goal in mind. Rather than attempting to revive preserved at brains, they favour brain emulation, which involves creating a high-resolution digital replica of a whole human brain. The idea is that a replica based on a person's detailed brain connectome data could recreate their memories – and even, perhaps, become conscious.
Nectome is focusing its efforts on developing “a process … to perfectly preserve every neuron and synapse of the brain… [to] maintain its quality, down to the nanoscale, for a hundred years or moreâ€. But rather than revive the brain itself, Nectome's founder claims that it will eventually be possible to digitise its structure, and the neural information stored within it, then upload the data to a supercomputer, to emulate the individual's mind and consciousness.
*
The study and mapping of the brain is called connectomics. It's a relatively recent neuroscientific field that aims to comprehensively map the pathways, circuits and synaptic connections in the nervous system. It requires looking at ultra-thin slices of brain tissue under powerful microscopes in order to reconstruct the fine structure of the cell bodies, fibres and synapses.
This approach is often combined with neuronal “tracing†methods that use genetic engineering to colourise individual cells so that their fibres and connections can be delineated more easily. At a larger scale, sophisticated imaging techniques are used to visualise the bundles of nerve fibres that serve as long-range connections between distant brain regions and investigate how different brain regions form their networks.
Connectomics research is extremely laborious and produces enormous data sets, which aren't always compatible with each other. The first organism to have its entire nervous system mapped was the millimetre-long worm C. elegans in 1984. This tiny creature's nervous system contains just 304 neurons; yet, with only electron microscopy at his disposal, it took biologist Sydney Brenner 10 years to map their layout and connections.
Fast-forward 40 years, to October 2024. The FlyWire Consortium reported that they had mapped the entire connectome of the fruit fly D. melanogaster. The fruit fly has a brain the size of a pinhead, but even with more sophisticated methods, it still took this very large team, spread over 127 academic institutions around the world, six years to map the connectome.
The human brain is not only vastly bigger than that of a fruit fly, but also vastly more complex. A 2024 study led by researchers at Harvard used electron microscopy to image and reconstruct a cubic millimetre-sized piece of human brain tissue that had been removed from a patient during surgery for epilepsy. The researchers cut the tissue into more than 5,000 sections and then imaged each one, yielding 1,400 terabytes of data.
They then generated a 3D reconstruction, revealing just how dense and intricate this tiny chunk was: it contained some 57,000 neurons and glial (or non-neuronal) cells. The entire human brain contains approximately 80 billion neurons and at least twice as many glial cells, which between them form hundreds of trillions of synapses.
*
If the brain could be preserved, then perhaps the information could be extracted at some point in the future. But preserving the fine structure of this extremely delicate and intricate tissue is no mean feat. Simply “fixing†a brain to prevent chemical changes and then freezing it leads to the formation of ice crystals that distort and damage the fine structure of the delicate tissue. Researchers at Nectome, the Cryonics Institute and other companies use a method called aldehyde-stabilised cryopreservation in order to prevent this.
Developed by Aurelia Song and Greg Fahy, this involves passing two solutions through the brain – first a solution of aldehydes, and then a highly-concentrated solution of cryoprotectant chemicals – to prevent ice formation during cooling and turn the entire system “into a solid glassâ€, as they wrote in their 2015 paper. They added that it “promises to be a powerful new technique in the quest of connectomics researchers to unravel the mysteries of the mindâ€.
Song went on to co-found Nectome, which, according to its website, is “dedicated to advancing the science of memory preservationâ€, so that it can “preserve people within minutes after legal death in order to preserve the body and brain to an extremely high level of qualityâ€. The company received a $960,000 federal grant from the US National Institute of Mental Health, a further $120,000 from the technology start-up accelerator Y Combinator, and was also awarded an $80,000 prize by the Brain Preservation Foundation.
Nectome quickly became mired in controversy, however, when it announced plans to begin the process on anaesthetized terminally ill patients, passing the cryopreservant chemicals through their brain while they are dying. Subsequently, MIT neuroscientist Ed Boyden stopped collaborating with Nectome, and the university terminated its subcontract with the company.
Song's latest study uses pig brains, whose anatomy and vasculature closely resemble those of humans, to show that her new protocol “results in connectomically traceable whole brains†that could be stored economically, and maintain stability, “even for thousands of yearsâ€. A draft paper was published in March but has not yet been peer reviewed; it concludes that the new protocol “is compatible with human brain preservation after physician-assisted death, provided that the blood washout is initiated less than about 14 min after cardiac arrestâ€. In practical terms,this means someone from Nectome would have to be onhand while the individual is euthanised, in order to begin the cryopreservation protocol within minutes of death.
At around the same time this study came out, Fahy and his colleagues published another draft paper demonstrating that the structure of whole rabbit brains “can be preserved by vitrification without prior aldehyde fixationâ€, by using a cryoprotectant called M22. This would further reduce damage to the brain tissue. The researchers then investigated the effects on tissue biopsies from the human cerebral cortex, showing that these had “no signs of ice crystal damageâ€. The process caused “brain shrinkage sufficient to distort and obscure neuroanatomical details†– but this was, they say, reversed upon rehydration.
The human brain tissue samples used in this study came from L. Stephen Cole, a gerontology researcher at UCLA's David Geffen School of Medicine with a strong interest incryonics. Cole died of pancreatic cancer in 2014 and had asked Fahy to perform the procedure on his brain. Soon after his death, Cole's body was transported to the Alcor Life Extension Foundation in Scottsdale, Arizona, whereupon his brain was removed and stored at -146 °C.
*
What about the future prospect of reviving these brains? There has been some progress in showing the potential for brain revivification, although none on human tissue. A study published this year by researchers in Germany shows that slices of rat brain tissue can recover their electrical activity after being rendered completely immobile in cryogenic glass by vitrification. However, this work, published in the Proceedings of the National Academy of Sciences, only demonstrates short-term recovery – the tissue underwent deterioration after 10-15 hours.
Lead author Alexander German still believes this was a significant step. He said that the study “showed intact synapses and dendrites … mitochondrial activity … neurons firing and … synapses communicatingâ€. He also said they managed to recover “the cellular mechanism for learning and memoryâ€. In other words, the tissue appeared to be fully, or at least mostly, functional. “We did not reanimate or revive a conscious brain,†he conceded, but didn't rule out the possibility of brain function recovery “in the distant futureâ€.
And then there is the digital route, which comes with its own challenges. In 2008, Anders Sandberg and Nick Bostrom of the Future of Humanity Institute at Oxford published a report entitled “Whole Brain Emulation: A Roadmapâ€, which contains the most detailed, and most widely cited, estimate of how much computing power would be needed to build and run a simulation of the human brain.
According to Sandberg and Bostrom, a low-resolution model of a brain that simulates neurons as simple “onoff†spiking units, and synapses as “weighted†connections with different strengths, would require between one petaflop and one exaflop of computing power. For comparison, the world's fastest supercomputer – at China's National Supercomputing Center – has achieved just under 2.2 exaflops. A more detailed model that incorporates gene expression, protein–protein interactions and other molecular level processes that are probably important for neuronal function would likely require at least 10 million exaflops of computing power.
Storing all this data would require at least an exabyte (one billion gigabytes), or possibly two or three. Although it's impossible to rank AI data centres on the basis of their storage capacity, one contender for the “biggest†is xAI's Colossus supercomputer in Memphis, Tennessee, which has an estimated 0.5 exabytes of total storage for training its algorithm, alongside some 200,000 graphics processing units.
*
Even when the rapid pace of technological advancement and the role of artificial intelligence in accelerating scientific discovery are factored in, whole brain emulation is still a long way off. According to another survey by the neuroscientist Zeleznikow-Johnston: “When predicting the future feasibility of whole brain emulation, the median participant estimated that this would be achieved for C. elegans [the millimetre-long worm] around 2045, mice around 2065, and humans around 2125.â€
Others believe that it may simply not be achievable. The possibility of brain emulation rests on several assumptions. First, it is assumed that learning occurs by the modification of synaptic connections, that memories are stored in sparsely distributed networks of neurons and their synapses, and that retrieval of a memory involves the “reactivation†of the neuronal network in which it is stored. There is strong evidence to support these claims. There is, however, much that we still do not know about the neural basis of learning and memory. Neurons and synapses are almost certainly parts of a bigger picture that we still do not see.
That the replica could become conscious seems even more far-fetched. Neuroscientists are still struggling to define consciousness, let alone understand its underlying brain mechanisms. And while there's little doubt that the brain plays an important role in generating our conscious experiences, it's far from clear that the brain is sufficient for human consciousness. The brain, body and environment form a complex, dynamic system, and consciousness emerges from the interactions between them.
All things considered, Sam Altman's assumption that his “brain will be uploaded to the cloud†is probably unrealistic. And if mapping his whole connectome then uploading it to a supercomputer does become possible at some point in the future, it would also be an extremely selfish and immoral thing to do. AI data centres and other supercomputers consume vast amounts of energy and water, so the endeavour would divert these resources from more important uses, such as sustaining human life.
As of 2026, just under 700 people have been cryopreserved. This number is increasing, although very slowly, with more people preserving only their brains, not their bodies, probably because of the lower cost of the procedure. Cryopreservation serves as a techno-religion that offers the hope of transcending biology to achieve a form of immortality. The idea is gaining popularity, with membership of cryonics organisations also steadily increasing.
Mind uploading will almost certainly be nothing more than a transhumanist fantasy, and even if the science does take us closer, there are ethical quandaries involved. Yet there may be some positives, and we are bound to learn more about how our brains function along the way.
