Can We Live Forever? Transcendence and Mind Uploading

In 2014 Johnny Depp played a scientist who dies and then... wakes up as a computer program. The film was called "Transcendence" and got pretty mixed reviews. Whatever the critics said, the question it asked is one of the most interesting you can ask about futuristic visions of immortality: can a human being be moved into a machine and cheat death that way?

A dozen or so years after the premiere, it looks like the world is slowly catching up with the script. Neuralink is implanting chips in people's brains. Artificial intelligence writes code, generates images and music. And in March 2026 a San Francisco startup announced that it had "uploaded" the entire brain of a fruit fly to a computer, where it started controlling a virtual body. The headlines said: the first brain upload in history.

So what's the deal? Is digital immortality a matter of a decade, a century, or will it never happen at all? I decided to look into it seriously: not at the level of TikTok reels from pseudoscientists, but by digging through scientific papers, industry reports, company documents and calculations that someone actually did, backed by data and science. This post is the result of that work, translated into language anyone can follow if they like technology but didn't study neurobiology.

Fair warning: the answer is neither "yes" nor "no". It's more interesting than you can imagine at this point.

The version for busy people

If you don't have time, here are the key takeaways:

No, today you can't move a human mind into a computer, and nobody is even close. Humanity's best result is a working digital fruit fly (about 140,000 neurons). You have 86 billion. That's a difference of roughly 600,000 times.
The biggest problem isn't computing power. There are three problems: we don't know exactly what needs to be copied to preserve a mind; we can't scan it in less than thousands of years; and even if we pulled it off, the result would be a copy, while you would die anyway.
Myth: "we're already implanting chips that replace part of the brain". Not true. The best implants read signals from around a thousand electrodes and replace no tissue at all. Reading and replacing are two completely different things.
The companies exist, from serious ones (Eon Systems, Neuralink, Merge Labs) to controversial ones (Nectome) to plain hoaxes (BrainBridge). None of them sells immortality. The ones that tried ended in scandal.
The most realistic path to a longer life is biology, not copying your brain onto a hard drive. And even that promises living "longer", not "forever".

If you want to know why, read on.

1. Let's start with the film, because that's where a very bold vision is laid out

A quick plot refresher. Will Caster (Johnny Depp) is a brilliant AI researcher building a system called PINN, short for Physically Independent Neural Network. He gets shot by technophobic terrorists from a group called RIFT with a polonium-laced bullet and is left with only a few weeks to live. His wife Evelyn (Rebecca Hall) and his friend Max (Paul Bettany) attach electrodes to his head, record his brain activity and "upload" it into PINN. Will dies, but moments later he speaks from a screen, as his own computer version. The rest of the film is about digital Will becoming superhumanly intelligent, taking over the internet, building nanobots and starting to change the world.

Remember the mechanism from the film, because we'll keep coming back to it: reading the brain from the outside with electrodes, in a living person, and transferring the data into an existing artificial intelligence.

That is exactly the part that is pure fiction. And here's the first irony of this whole story: the only method science actually has for reading enough detail out of a brain works exclusively on a dead brain, chemically fixed and cut into slices thinner than a thousandth of a hair. We'll get to that.

2. Three different things we throw into one bag labelled "immortality"

Before we go further, it's worth sorting out the terms, because the media (and company marketing) mix up several completely different ideas.

Idea one: uploading, i.e. emulating the whole brain. In English it's whole brain emulation, or WBE for short. The point is to scan a specific person's brain in such detail that it can be recreated in a computer and switched on. Not "an artificial intelligence similar to you", but a copy of your brain, neuron by neuron. That's the "Transcendence" scenario.

Idea two: moving the biological brain into another body. In other words, a head or brain transplant. Your brain stays biological; only the "casing", meaning the body, changes. It sounds like a B-movie horror, but there have been doctors who announced they would do it.

Idea three: extending biological life. So-called longevity: reprogramming cells, rejuvenating tissue, treating ageing and various diseases. It isn't immortality, just "longer and healthier". And that's exactly where the biggest money flows.

There's also a fourth idea, which is not any of the above but is sometimes sold as if it were: the digital twin. A chatbot trained on your emails, recordings and photos that, after your death, talks to your family "in your voice". It's an imitation of certain character and personality traits, not the real you. We'll come back to it, because it's the most common con in this space.

This piece is mainly about idea one, because that's what stirs up the strongest emotions when scientists talk about "eternal life in a computer".

3. Brain basics: what our "self" is made of

To understand why copying a brain is so hard, you need to know what exactly we'd be copying. As a layman myself, I promise to explain things in the simplest, most understandable way.

Neurons. The basic cells of the brain, a bit like the transistors in a processor, only alive and far more complicated. You have about 86 billion of them. For comparison: the Milky Way has roughly 100–400 billion stars, so you are more or less a third of a galaxy.

Synapses. The places where neurons connect with each other and pass signals along. A single neuron can have thousands of them. The whole brain has between 100 and 1,000 trillion (that's thousands of times more than there are stars in the Milky Way). It's in the synapses, in their strength and arrangement, that what we call memory, habits and character live.

The connectome. The complete map: which neurons connect to which. Imagine a city plan with every street, every junction and every cycle path marked on it. Making a map like that for the human brain is the Holy Grail of modern neurobiology and, as you'll see, we're very far from it.

And now an important thing the media usually keep quiet about. A street map is not the same thing as a city. A city plan doesn't show how many cars drive down which street, where the traffic jams are, how the traffic lights work, or what the traffic rules are. In the brain, the equivalents of those things are:

synaptic strength (one connection can be barely a whisper, a gentle suggestion, while another is a shout, an overriding compulsion that steers us, and this changes from minute to minute as you learn),
chemistry: neurotransmitters and hormones that globally "retune" entire regions of the brain (which is why you're a different person after coffee, a different one after a sleepless night, and behave differently yet again when you've fallen in love),
glial cells, which for decades were considered the "glue" holding neurons in place, and which we now know take part in information processing; in the examined fragment of the human cortex there were twice as many of them as neurons,
computation inside a single cell: an individual neuron isn't a simple switch, its "branches" (dendrites) perform local computations of their own,
dynamic state: what's happening in the brain right now, not just its static structure.

The key question is: at what level of detail does a brain have to be copied for the copy to really be "you"? Just the connections? Connections plus strength? Plus chemistry? Plus the state of every molecule? Nobody knows. And that's not a technical detail, it's the foundation of the whole discussion. The most famous document in the field, the whole brain emulation roadmap by Anders Sandberg and Nick Bostrom of Oxford (2008), estimates the required computing power at anywhere from 10¹⁵ to 10²⁵ operations per second. That spread is ten orders of magnitude. It's as if someone asked how much a car costs and the answer was "somewhere between $10,000 and $100 trillion, depending on what a car even is".

4. How much data is that? The numbers that frame these hypothetical musings

This is where it gets concrete, and this is where most of the optimism dies.

One cubic millimetre of human brain

In May 2024 a team from Harvard and Google published in Science the most detailed map of a piece of human brain ever made. The sample came from a 45-year-old woman who had surgery for epilepsy, and it had a volume of one cubic millimetre, roughly a grain of sand. Inside they found about 57,000 cells, about 150 million synapses and 230 millimetres of blood vessels.

The raw data from this "grain of sand" took up 1.4 petabytes. That's 1.4 million gigabytes, or roughly 300,000 movies in HD quality. It's worth stressing that we're talking about one cubic millimetre here.

The human brain has about 1.2–1.4 million cubic millimetres. According to the 2025 State of Brain Emulation report, the examined fragment is about 0.00007% of the whole. And one more thing: of the roughly 16,000 cells identified in this sample, only 104 neurons were fully verified by hand. The rest is an automatic reconstruction with errors that nobody has corrected yet.

The whole brain

Google says it outright in its post: a complete map of the human brain would require collecting a zettabyte of data, i.e. a billion terabytes. A more detailed calculation was done by Collins, Huffman and Koene in their paper on methods for imaging whole mammalian brains: at a resolution of 18 nanometres, the raw images of a mouse brain come to about 86 petabytes, and of a human brain about 206 exabytes. The hard drives alone to store those images would cost about $2.26 billion at 2023 prices. If you stacked those drives one on top of another (206 million 1 TB drives, each about 2.6 cm thick), the pile would be over 5,000 kilometres high.

To be fair: those are raw images. A finished, processed map would be much smaller. But to have it, you first have to take those pictures.

How these pictures are taken (and why it can't be done to a living person)

The only method that gives synapse-level resolution is serial electron microscopy. It goes like this:

The brain has to be chemically fixed (to put it bluntly, embalmed, before the tissue starts to decompose).
It's cut into slices about 30 nanometres thick. A human hair is about 70,000 nanometres across, so we're talking about slices more than two thousand times thinner than a hair.
Each slice is photographed with an electron microscope.
Billions of images are stitched together into a three-dimensional block, and artificial intelligence is told to trace every neuron through it.
The AI's mistakes are corrected by hand. This stage, so-called proofreading, is estimated to account for over 95% of the total cost. Checking the fruit fly brain map took people a combined 33 years of work (the sum of the working hours of all team members).

Note points one and two. This method is destructive and post-mortem by definition. There's no version where "we scan a living Will Caster with electrodes". MRI, the only non-invasive way of looking at the brain, sees detail on the order of a millimetre; synapses are tens of nanometres. That's a difference of tens of thousands of times along each of the three axes. Scanning a brain without destroying it doesn't exist, and we don't even have an idea of what it might look like.

How long it would take

And here we hit the hardest wall. At today's microscope throughput, Collins and co-authors estimate that scanning a single human brain on a single multi-beam microscope would take something on the order of millions of years. A fleet of four hundred microscopes costing $100 million would cut that to about 9,000 years. Another calculation, from a report on neuroscience for AI safety, says that without further acceleration of the technology you'd need about 50,000 microscopes working in parallel.

For scale: mapping just a mouse brain (500 times bigger than the earlier human sample, but 1,000 times smaller than a whole human brain) was estimated at about $10 billion and up to 17 years of work. The NIH BRAIN CONNECTS programme is currently funding an attempt to map 10 cubic millimetres of mouse hippocampus, i.e. 2–3% of a mouse brain, for about $33 million.

Does that mean we'll never manage it? Not quite. New methods are emerging: expansion microscopy (physically "inflating" the tissue so it can be viewed under an ordinary light microscope), X-ray tomography at synchrotrons, and cheaper approaches developed by organisations such as E11 Bio, which is aiming for a hundredfold reduction in cost. Optimistic reports say that with massive parallel work and automation the time could be cut to a few years. But that's still in "what if" territory.

5. The lesson of the worm: we've had the full map since 1986 and it still doesn't work

In my opinion this is the most important part of the whole piece, because it shows that the problem doesn't lie (only) in computing power.

There's a nematode, Caenorhabditis elegans, about a millimetre long. It has exactly 302 neurons and about 7,000 synapses. In 1986, after a decade of slicing and photographing, John White and Sydney Brenner's team published its complete connectome. The paper goes by the unofficial title "The Mind of a Worm". For forty years we've had a complete map of every connection in this brain.

And for forty years nobody has built a working simulation of this worm that the scientific community would accept as an emulation. The OpenWorm project has been trying since 2011. A 2021 summary is titled "Whole Brain Emulation: No Progress on C. elegans After 10 Years". There are also optimistic voices saying that it can be done and that we should do it, but even they admit that we lack data on how each connection works.

Why isn't the map enough? Because the map tells you that neuron A connects to neuron B. It doesn't tell you how strong that connection is, how it changes over time, which neurotransmitters it uses, how it responds to hormones, or how neuron A computes signals internally. Neurobiologist Michael Hendricks of McGill University, who studies this very worm himself, called the promises of uploading "false hope" in MIT Technology Review for exactly this reason: the connectome is necessary, but it isn't sufficient to recreate a nervous system.

The conclusion is brutal and simple. If we can't get 302 neurons running, neurons we've had the map of since 1986, then every claim about running 86 billion is marketing, not science.

And now, to be fair, the other side of the coin, because in 2024 something happened that gives hope and slightly calls the arguments above into question.

6. The breakthrough: a fruit fly that "works" in a computer (2024–2026)

Step 1: the map (October 2024)

The FlyWire consortium, led by Princeton University and bringing together 146 labs from 122 institutions, published in Nature the first complete map of an adult fruit fly brain: 139,255 neurons and over 50 million synapses. A fly's brain is the size of a poppy seed. To this day it's the largest complete connectome of an adult animal. Anyone can browse the map at flywire.ai, and ordinary people helped correct it too, much like playtesting a game.

Step 2: a model that predicts behaviour (2024)

Philip Shiu from Berkeley and his team built a computational model of the entire fly brain on top of this map. They gave every neuron the simplest possible formula ("collect signals; when you cross a threshold, fire"). There was no machine learning involved, no training. And it turned out that the model predicts which motor neurons will respond to a given stimulus (e.g. sugar on the tongue) with about 95% accuracy. What's more, this whole brain can be run on a laptop.

Step 3: the body (7 March 2026)

San Francisco startup Eon Systems, where Shiu is now chief scientist, connected this digital brain to a virtual fly body in a physics simulator. The company's co-founder Alex Wissner-Gross described it on Substack as the first-ever "embodied whole-brain emulation that produces multiple behaviours". The virtual fly walks, grooms itself, drinks from a little dish, reacts to light and looks for sugar, and its behaviour matches that of a living insect about 91% of the time. When the researchers deliberately scrambled the wiring, accuracy dropped to 1%: meaning the behaviour really does come from the map and not from chance.

Eon Systems' advisory board includes, among others, George Church from Harvard, Stephen Wolfram and the aforementioned Anders Sandberg. According to unofficial reports, the company raised about $3.5 million in seed funding (including from Protocol Labs and Larry Page's family office). The plan: a mouse brain (70 million neurons) within two years, a human brain "in decades".

What this really means

This is real science and a real milestone. But before anyone pops the champagne, a few caveats.

First, scale. Going from 139,000 neurons to 86 billion is about 600,000 times. It's as if, after simulating traffic in one small town, you announced that you'd soon simulate traffic in every city on Earth (and on a few thousand other planets too), and that's assuming cities contain nothing but streets.

Second, simplification. The model uses switch-like neurons. That works for a fly, because its behaviours are largely "genetically programmed" and rigidly "hard-wired". Will that kind of simplicity be enough for the mammalian cerebral cortex, where everything depends on learning, chemistry and plasticity? Nobody knows. The authors themselves say so openly.

Third, it isn't a specific fly. It's an emulation of "a fly in general", built on the map of one individual, but without its memories, without its state, without anything that would distinguish it from any other fly. Even if we had a map of your brain, the question "is this you" remains open. More on that in section 10.

What about mammals?

In April 2025 the MICrONS project (Princeton, the Allen Institute, Baylor) published in Nature a package of eight papers on one cubic millimetre of mouse visual cortex: over a petabyte of data, about 200,000 cells, about 120,000 neurons and 523 million synapses, plus, for the first time at this scale, recorded activity from 75,000 of those neurons in a living mouse watching videos. Nature Methods named electron-microscopy connectomics its "Method of the Year 2025". It's a grain of sand from a mouse brain. But it's the best picture of the mammalian cortex we have.

7. Who's working on this (and for how much)

Here's the overview I was looking for at the start and couldn't find anywhere in one place. I've split the companies into "five worlds", because marketing deliberately blurs them together, and they're completely different things.

World A: brain emulation

Eon Systems (San Francisco, 2025) is the only commercial company doing literally what this post is about. I described it above. The Carboncopies Foundation (Randal Koene) is a non-profit that has for years been coordinating researchers interested in emulation; it doesn't build a product.

World B: brain mapping

Google Research with the Lichtman lab at Harvard (the human sample), FlyWire (the fly), MICrONS (the mouse), the Allen Institute, E11 Bio (cheaper tools), and above it all the US BRAIN Initiative, which funds most of this work. This is public science, not business.

World C: brain preservation and cryonics

This is where the part you need to be careful with begins.

Nectome. In 2018 MIT graduate Robert McIntyre got into the most famous startup accelerator, Y Combinator, with a company that had won a prize for preserving a large mammal's brain and received a government grant. Its method, so-called aldehyde-stabilized cryopreservation, involves pumping a fixative (glutaraldehyde) into the brain and freezing it in a state ideal for future scanning. The catch: for it to work, the brain has to be fresh. Which means, as MIT Technology Review wrote, the procedure has to be performed on a living person under anaesthesia, and it is "100 percent fatal". The company considered piggybacking on California's assisted-dying law for the terminally ill. It collected over $200,000 in deposits ($10,000 per person; the list included Sam Altman). After the media storm, MIT cut ties with it, and neurobiologist Sten Linnarsson of the Karolinska Institutet said the whole company rested on a false premise. Nectome still exists as a research company, but this episode defines how investors and scientists react to the words "brain upload".

Cryonics. Companies such as Alcor (Arizona, since 1972, about 250 "patients" in liquid nitrogen tanks), the Cryonics Institute (Michigan, since 1976, about 270) and Europe's Tomorrow Bio (Berlin, since 2020, about 20 patients, raised €5 million in 2025 to expand into the US). Prices: from about $28,000 at the Cryonics Institute, through $80,000 for just the head at Alcor, to about $200,000 for the whole body. Globally, about 500–650 people are in cryonic suspension.

And here's the important bit. Cryonics and chemical preservation are not uploading. They're a bet: "we'll freeze you, and someone in the future will figure out how to read you or bring you back". Nobody has ever revived or read a preserved human brain. Critics (like the aforementioned Hendricks) argue that current freezing methods destroy exactly the details that would be needed. Supporters argue that enough is preserved. There's no resolution, because there's no technology to read it. What's being sold right now is hope.

World D: brain–computer interfaces (BCI)

This is the only one of these worlds where real products, patients and real money exist.

Neuralink (Elon Musk, since 2016). The N1 chip has 1,024 electrodes on 64 threads thinner than a hair, inserted into the motor cortex by a surgical robot. The first patient received the implant in January 2024; by the end of 2025 there were a dozen or so across four countries (the US, Canada, the UK and the United Arab Emirates). Paralysed patients control a cursor, play games and operate a robotic arm. In June 2025 the company raised $650 million at a valuation of about $9 billion; in total it has over $1.3 billion from investors (and, according to reports of another round in 2026, close to $1.9 billion).

Merge Labs (Sam Altman, revealed in January 2026). It raised about $252 million in a seed round led by OpenAI, at a valuation of about $850 million. A non-invasive approach: ultrasound and molecules instead of electrodes stuck into the brain. The founders themselves say it's a project for "decades, not years".

Then there's Synchron (an electrode inserted through a blood vessel, without opening the skull; its investors include Bezos and Gates), Precision Neuroscience, Blackrock Neurotech, Paradromics, Science Corp (Max Hodak, Neuralink's former president) and Kernel.

Remember one sentence about this whole world: these devices read signals from a small number of neurons or stimulate them. None of them replaces brain tissue. I'll expand on that in section 8.

World E: computers made of living neurons

Since 2025, Melbourne-based Cortical Labs has been selling the CL1, a "biological computer": about 800,000 living human neurons, grown from donors' skin or blood cells, growing on a silicon chip. The neurons live for up to six months, and the price is about $35,000 per unit, or $300 a week in the cloud ("wetware as a service"). The device's predecessor learned to play Pong in 2022. Switzerland's FinalSpark rents out brain organoids in a similar way.

It's fascinating, but it has nothing to do with transferring a mind. It's growing anonymous neurons for computation. Nobody claims that "someone is in there" in those dishes.

Big projects that have ended

It's worth knowing that the two biggest brain simulation programmes in history have already shut down.

The Blue Brain Project (EPFL, Switzerland, 2005–2024), led by Henry Markram, built detailed simulations of fragments of rat cortex (about 30,000 neurons in 2015) and an entire methodology of "simulation neuroscience". It wound up in December 2024.

The Human Brain Project (EU, 2013–2023) cost about €607 million, brought together 155 institutions and produced over 3,000 publications. It left behind the EBRAINS infrastructure, brain atlases and neuromorphic chips. It did not create a digital human brain, even though that's exactly how it was advertised in 2013, and that was one of the main reasons the project was criticised.

Hoaxes and ghost projects

The 2045 Initiative of Russian billionaire Dmitry Itskov (2011) promised the "Avatar" plan: a thought-controlled robot by 2020, a brain transplant into an artificial body by 2025, an artificial brain by 2035 and hologram "immortality" in 2045. Not a single deadline was met, and the project's website hasn't been updated in several years.

BrainBridge (2024). Remember the viral video of "the world's first head transplant machine", with robot surgeons and a futuristic operating theatre? Tens of millions of people watched it. As MIT Technology Review established, it was a computer animation created by Hashem Al-Ghaili, the creator of an earlier fake, "EctoLife", about artificial wombs. The company BrainBridge doesn't exist in any registry.

Digital twins

HereAfter AI, StoryFile, Eternos, Replika-style avatars. You record yourself, upload your emails and photos, and after your death your family talks to a bot that speaks in your voice and your style. As a keepsake, as a way of saying goodbye, it can have value. But remember: this bot doesn't have your memories, your consciousness, or anything that is you. It's a statistical imitation of style. Any company that calls it "life after death" or "digital immortality" is misleading you.

Everything in one table

Who	What they actually do	Money	Stage	How far from "Transcendence"
Eon Systems	Connectome emulation (fly, then mouse)	approx. $3.5M (unofficial)	Insect demo (March 2026)	Enormously far: 600,000 times fewer neurons than a human
FlyWire / MICrONS / H01	Brain maps (fly, 1 mm³ of mouse, 1 mm³ of human)	Public grants, Google	Science	Mapping only; the human sample is 0.00007% of the brain
E11 Bio	Cheaper mapping methods	Philanthropy	Research	Tools, not a mind
Nectome	Chemical brain preservation	Over $200K in deposits, a grant	Research	No readout method; the procedure is fatal
Alcor / CI / Tomorrow Bio	Post-mortem cryonics	Fees; Tomorrow: €5M	Service	A bet on the future; 0 revivals in history
Neuralink	Signal-reading implant (1,024 electrodes)	Over $1.3B	Clinical trials	Reading is not uploading
Merge Labs	Non-invasive interface (ultrasound)	approx. $252M	Lab	"Decades, not years" (their own words)
Cortical Labs / FinalSpark	Computing on living neurons	Product from $35K	On sale	Not applicable: nobody is being transferred
2045 Initiative	The "Avatar" vision	Private	Dead	Didn't meet a single deadline
BrainBridge	CGI video	0	Fake	Not a company
HereAfter / StoryFile / Eternos	Chatbot imitating a person	Varies	Product	It's an imitation, not you

8. The myth of chips that "replace part of the brain"

In discussions about brain uploading, one argument keeps coming up: "they're already implanting chips that replace a piece of the brain, it's just a matter of scale". I've even come across a version claiming that implants "replace 20% of the brain". That's not true, and it's worth understanding why, because the difference is fundamental.

What Neuralink does. The N1 chip has 1,024 electrodes. The brain has 86 billion neurons. The electrodes "eavesdrop" on electrical activity in a small patch of the motor cortex, and an algorithm learns which pattern means "move the cursor left". It's brilliant engineering and it changes the lives of paralysed people. But it's reading, like a microphone held against the wall of a skyscraper that picks up a fragment of one conversation. Nothing in that skyscraper has been replaced.

What cochlear and retinal implants do. They replace a damaged sense organ (ear, eye) by delivering a signal to the brain. The brain still does all the work. It's replacing a cable, not the processor.

The hippocampal prosthesis (Theodore Berger, Robert Hampson). This is the closest thing to "replacing a brain function" that exists. The device records activity patterns in the hippocampus (the memory centre) during memorisation and plays them back, strengthening the memory trace. In epilepsy patients it improved memory test scores by anywhere from a dozen or so to several dozen percent. But it assists a working hippocampus rather than replacing it, and it works in a dozen or so patients under experimental conditions.

Speech decoding. Since 2023, implants from Stanford and UCSF have allowed people who lost the ability to speak to "talk" through a computer at a rate of several dozen words per minute. Again: reading intent, not replacing anything.

Why this matters for the immortality question

There's one concept of uploading that theoretically sidesteps the "copy or original" problem (more on that in a moment). Hans Moravec proposed it in 1988: gradual replacement. You replace the brain piece by piece with artificial components that do exactly the same thing, until one day you're entirely artificial, and there was never a moment when you "died". The Ship of Theseus, but applied to your head.

For that to be possible, you'd need to be able to replace a piece of cortex so that the rest of the brain doesn't notice the difference. And that requires: (1) an interface millions of times denser than today's, (2) an understanding of how that piece computes, i.e. solving the problem from section 3, and (3) an artificial component that can learn and change like living tissue. We have none of those three things. Merge Labs, which is aiming precisely at a denser, non-invasive interface, honestly talks about decades. The gap between "we read 1,024 electrodes" and "we replace tissue" isn't a matter of scale, it's a matter of kind.

9. What about a head transplant?

If you can't move a mind into a computer, maybe you can move the brain into a new, younger body?

In 1970 neurosurgeon Robert White transplanted a monkey's head onto another monkey's body. The animal lived for a few days and was paralysed from the neck down, because a severed spinal cord can't be reconnected. In 2015 Italian neurosurgeon Sergio Canavero announced that within two years he would perform the first human head transplant, together with Chinese surgeon Ren Xiaoping. In 2017 they reported a "trial" on cadavers. There was never a living patient. The medical community deemed the whole thing unethical and scientifically unjustified.

Two barriers are fundamental. The spinal cord: once severed, it doesn't heal back together; Canavero promised that polyethylene glycol would help, but there's no convincing evidence for that. Immune rejection: the donor body will attack the foreign head (and vice versa), and the brain is exceptionally sensitive to inflammation.

The 2024 BrainBridge video I wrote about above was a fake. Verdict: a head transplant is no closer than uploading. It's just a different dead end. And even if it worked one day, it wouldn't give you immortality, just a second body for the same ageing brain.

10. The biggest problem: is the copy still you?

Let's assume for a moment that everything from the previous sections has been solved. We have the scanner, we have the computer, we have a complete understanding of the brain. We scan you and switch you on. Someone appears on the screen who has your memories, your sense of humour, your fears, and is absolutely convinced they're you.

The question: are you in there?

This problem has a long history in philosophy. Stanisław Lem described it in "Dialogues" (1957): if we break a person down into atoms and recreate them perfectly somewhere else, is it the same person, or a new one who merely thinks so? Derek Parfit in "Reasons and Persons" (1984) called it the teletransportation paradox. The key variant is the one where the teleporter doesn't destroy the original: you end up with two people, both convinced they're you. Which one is "real"? And if both, what does that even mean?

Let's apply this to brain uploading. The scan (reminder: destructive, post-mortem) creates a copy. From the copy's perspective, everything went fine: it fell asleep as a human and woke up in a computer. From your perspective, the first-person one, the one reading this text right now, nothing happened except that you died. MIT Technology Review put it gently in its piece on Nectome: someone very similar to you, though not exactly you, would wake up in a server.

There are a few ways to respond to this:

"Identity is psychological continuity, not specific matter" (Parfit). If the copy has your memories and personality, then it is you in the only sense that matters. Many transhumanists accept this answer. But it's a philosophical decision, not a scientific discovery, and it still doesn't solve the problem of two copies existing at once.
"Let's do it gradually" (Moravec). If we replace the brain piece by piece, there's never a moment of death and a jump. It's the only path that seems to preserve continuity. Except that, as I wrote in section 8, technologically it's the furthest away of all, and still nobody can prove that continuity of function equals continuity of the subject.
"We don't know, and there's no way to check." In my opinion, that's the only honest answer for now.

Does the copy feel anything at all?

There's an even deeper layer. Even if we accept the copy as "you", there's still the question of whether it experiences anything. David Chalmers called this "the hard problem of consciousness": we can explain how the brain processes information, but we don't know why subjective experience accompanies it.

One of the most influential contemporary theories of consciousness, Giulio Tononi's integrated information theory, says outright that an ordinary digital computer, even one perfectly simulating a brain, would have consciousness close to zero, because physical architecture matters, not just behaviour. Philosopher John Searle argued something similar: consciousness is a biological phenomenon, like digestion, and a simulation of digestion won't feed anyone. Others (functionalists) believe the substrate doesn't matter, only the organisation does. The dispute has been going on for decades, and there's no experiment that could settle it.

The consequence is unsettling: an emulation could be a so-called philosophical zombie. It behaves like you, talks like you, claims to feel, and yet "inside" there's nobody home.

How companies answer this problem

For the most part, they don't. Eon Systems talks about "emulation", not about "you". Nectome sold a "mind backup" without specifying who would wake up from it. Digital twin companies simply pretend the problem doesn't exist. It's a deliberate gap, because the honest answer ("we'll make a copy, and you'll die") spoils the pitch.

11. Law and ethics: to "upload yourself somewhere", you first have to die

Let's assume you're ready to accept that a copy will wake up in the computer. You still have a problem: the law.

Preservation requires death. Nectome's method only works on a fresh brain, which means it has to be performed on a living person under anaesthesia, which amounts to killing them. In the US, the company considered relying on the assisted-dying laws for the terminally ill that exist in several states. In Poland, euthanasia is a crime (Article 150 of the Penal Code). The same goes for most countries in the world.

Cryonics only after death. That's why cryonics companies maintain "standby" teams that wait at the bedside of dying clients so they can start the procedure seconds after death is pronounced. Every minute without oxygen destroys the brain, and the law doesn't allow them to start any earlier.

Implants are medical devices. Every brain–computer interface falls under the US FDA or the EU Medical Device Regulation (MDR). That means years of clinical trials and hundreds of millions of dollars before anything makes it beyond the experimental stage. Neuralink got a fast track from the FDA ("Breakthrough Device"), but it's still in the research phase, not on sale.

Neurorights, a new field of law. Chile became the first country to write protection of "brain activity" into its constitution, and in 2023 its Supreme Court ordered the company Emotiv to delete data from a certain senator's EEG headband. In the US, more and more states (Colorado since 2024, California since 2025, Montana, Connecticut) classify neural data as sensitive data. In November 2025, at a conference in Samarkand, UNESCO adopted the first global framework on the ethics of neurotechnology: non-binding, but establishing "mental privacy" as a principle. In the European Union, the AI Act bans subliminal techniques that manipulate people, and the GDPR covers health data. For a company wanting to work on human brains, it's an ever-denser minefield.

A digital person doesn't exist in law. If an emulation of a human were switched on tomorrow, it wouldn't have legal personhood. It couldn't have a bank account, inherit property "from itself", sign contracts, vote, or demand not to be switched off. It could be copied, sold, modified or deleted, and formally it wouldn't even be a crime. The whole area is a legal void.

12. Who pays for the "immortality startup"

Following the money tells you more about this industry than following the publications.

Pattern one: billionaires fund their own dreams. Neuralink is Musk's money (plus the funds that came in once there were patients). Altos Labs, the biggest longevity startup in history, got $3 billion in 2022 from Jeff Bezos and Yuri Milner. Retro Biosciences got $180 million from Sam Altman out of his own pocket, and Merge Labs $252 million from OpenAI, which Altman runs. The common denominator: the investor is both the source of capital and the name that lends the project credibility.

Pattern two: money flows where there's a medical pathway. Brain–computer interfaces raise hundreds of millions because they have a real market (paralysis, loss of speech, blindness), a real regulatory pathway and real patients. Brain emulation has none of those things. Eon Systems, the most serious company in this space, reportedly raised $3.5 million. Neuralink raised $650 million in a single round. That says everything about how investors value "uploading" versus "interfaces".

Pattern three: public science does the heavy lifting. The maps of the fly, the mouse and the human sample came from grants: the US BRAIN Initiative (in recent years a budget of around $300–400 million a year, and trending downwards), Europe's Horizon Europe, private foundations. Companies build on what someone else previously paid for with taxes.

Pattern four: reputation is the currency. Nectome showed what happens when you collect deposits for uploading: the media, a scandal, a severed partnership with MIT. BrainBridge showed that an immortality video goes viral and gets exposed just as fast. The 2045 Initiative showed what promises with dates look like once the dates have passed. Serious investors often keep well away from the word "immortality".

If someone wanted to build a company on this

I've thought this through, because the question "what if I gave it a go" comes naturally to anyone who likes building things. The honest picture looks like this.

A company selling uploading as a product can't currently be funded by anyone who isn't a billionaire or doesn't have a team with PhDs and a track record in connectomics. Not because investors are cowardly, but because there's currently no real product, there isn't even the underlying science, and there's a high reputational risk if it fails.

On the other hand, there are real, honest businesses around this topic where programming and product skills count, not a lab coat:

Tools for connectomics. A reminder: manually correcting the maps is over 95% of the cost. Any AI tool that cuts that cost has customers in labs around the world. It's a market of small but very motivated buyers. (Fun fact: you can join the research yourself by helping correct the fly map through FlyWire; it works like a game.)
The software layer for brain–computer interfaces. Neuralink and its competitors build the hardware. Apps, patient interfaces, decoders and tools for clinics are a separate, growing niche.
Honest "digital legacy". Products for recording your story, voice and memories for your family are fine, as long as nobody claims it's you living on.

Poland has its own research hubs too: the Nencki Institute in Warsaw (neurobiology, brain imaging), as well as small neurotech companies such as Gdańsk-based BrainScan (AI for analysing brain images) and Lublin-based Cortivision (mobile brain imaging). European programmes like the EIC Accelerator fund deep tech with amounts on the order of a few million euros. That's not money for immortality. It's money for tools that might, someday, bring us closer to an answer.

13. Film versus reality, point by point

Scene from "Transcendence"	Reality
Electrodes on living Will's head read his mind	Fiction. Electrodes (even implanted ones) read thousands of neurons out of 86 billion. The only method with sufficient resolution requires embalming the brain and slicing it up.
Will's mind "uploaded" into an existing AI (PINN)	Fiction. There's no way to "merge" brain data with a neural network. Real emulation is a separate simulation, neuron by neuron.
Will in the computer is still Will	Unresolved and probably false. It would be a copy; the original dies. The film itself plays on this uncertainty, too.
Digital Will becomes superhumanly intelligent in weeks	Fiction. A brain emulation, if it worked, would be roughly as smart as the original. Speeding it up is a separate, speculative topic.
Nanobots, taking over the network, regenerating bodies	Fiction, all of it.
Reading brain activity with electrodes	Real, but narrow: a cursor, speech, a robotic arm for paralysed people.
Brain map and simulation	Real at the level of a fly (2024–2026) and a grain of sand from a mouse brain.
"Merging" humans with AI	A real research direction (Merge Labs, Neuralink), but as an interface for control, not as a transfer of a person.

The biggest irony: the film shows uploading as a recording of a living person. The only currently realistic route is the exact opposite: death first, then slicing, then the brain map.

14. The verdict: when?

Let's pull it all together. Here are the obstacles, from the biggest down:

We don't know what to copy. The lesson of the worm. Without understanding how the connections work, not just where they are, the map is like a musical score when you can't read notes. It's a knowledge barrier, not a technical one.
The problem of the copy and of consciousness. Even a working upload produces a copy. And whether the copy feels anything, we have no way of checking.
Scanning. Thousands of years on today's hardware, and no non-destructive method.
Data and computation. Hundreds of exabytes of raw images; computing power unknown to within ten orders of magnitude.
Law and ethics. You have to die; a digital person has no rights.
Money and reputation. Funded by billionaires and grants; the word "immortality" often scares serious people off.

And here are the scenarios, labelled as fact, speculation or marketing:

Scenario	Status in 2026	Realistic horizon	Label
Destructive scan + human emulation, "it's still you"	Works for a fly. For a human: no scanner, no knowledge, the copy problem	If at all, many decades; the "it's still you" version perhaps never	Speculation (technology), marketing ("it's you")
Uploading without destroying the brain, while alive	No technological path exists	Science fiction	Fiction
Gradual replacement of the brain with artificial components	We can read from ~1,000 electrodes; we'd need millions of times more, plus an understanding of the neural code	Decades and more	Speculation
Head / brain transplant	Spinal cord doesn't heal, immune rejection, no living patient in history	No closer than uploading	Speculation, partly hoaxes
Digital twin (chatbot)	Works today	Already here	Fact (as a keepsake), marketing (as "life after death")
Cryonics / chemical preservation	Preserves tissue; nobody has ever revived or read anyone	A bet on an unknown future	Fact (preservation), speculation (revival)
Biological life extension	Billions of dollars, real research, first therapies in trials	Years and decades for "longer", not "forever"	Fact (research), speculation (size of the effect)

What do the forecasts say? In 2008 Sandberg and Bostrom considered emulation achievable by extrapolating existing technologies, but under many conditions and on a mid-century horizon. Ray Kurzweil has been betting on 2045 for years. Eon Systems says "a mouse in two years, a human in decades". Critics, like Brazilian neuroscientist Miguel Nicolelis, argue that the brain can't be reduced to a digital medium at all. The view from the AI community is interesting too: most researchers believe that AI will match humans long before we manage to emulate a brain, and that scanning alone is currently about 100,000 times too slow.

My answer to the question in the title: no, you can't currently live forever, and brain uploading is unlikely to change that in the coming years. Real progress is happening, but in brain science, not in immortality. The fly in the computer is a fascinating result that could change medicine, neurology and artificial intelligence. But it isn't the first step towards your eternal life, just as the Wright brothers' first flight wasn't the first step towards teleportation.

15. What this means for you

A few practical takeaways to finish with.

Don't pay deposits for an upload. If anyone offers to "save your mind" for money, they're selling hope without any technology to read it. Cryonics is a different category (legal, open about the uncertainty), but it's still a kind of hypothetical bet, not a service.

A digital twin is not you. If you want to leave your family recordings, stories and your voice, that's a lovely idea. Just don't let anyone convince you it's a form of life after death.

If you're a developer, this is a real niche. Connectomics is drowning in data and errors that need fixing. AI tools for reconstructing, visualising and verifying brain maps are work that someone has to do, and that can be done from an ordinary laptop. You can start with FlyWire, where correcting the map works like a game.

What to watch. Whether Eon Systems delivers on its mouse promise (70 million neurons) on schedule. Whether E11 Bio actually cuts the cost of mapping a hundredfold. What Merge Labs shows in five years. Whether someone finally gets that wretched worm running.

Be sceptical of videos. BrainBridge taught us that in the era of generative AI, the first "head transplant" or "human upload" will show up as a fake first, and it'll get more views than the real science.

Finally, back to the film. "Transcendence" ends ambiguously: we don't know whether it was Will in the machine, or something pretending to be him. The screenwriters didn't have an answer. Science doesn't have one either, but at least today it can say precisely why. And the conclusion from that precision may be less spectacular than in the cinema, but it's more useful: the life you have is the only one you can reasonably count on today. It's worth living it, rather than waiting for a backup.

Glossary

Neuron – a nerve cell; the basic processing unit of the brain. You have about 86 billion of them.
Synapse – a connection between neurons through which a signal passes. The brain has hundreds of trillions of them.
Connectome – a complete map of the connections between neurons. We've had one for the worm since 1986, for the fly since 2024; for humans we have 0.00007%.
Whole brain emulation (WBE, upload) – recreating a specific brain in a computer and running it.
Proofreading – manually correcting the errors AI makes when reconstructing a brain map; over 95% of the cost of connectomics.
Serial electron microscopy – the only method that can see synapses; it requires fixing the brain and cutting it into slices about 30 nanometres thick.
BCI (brain–computer interface) – a device that reads or stimulates brain activity, e.g. Neuralink. It reads; it doesn't replace.
Cryonics – storing bodies or brains in liquid nitrogen after death in the hope of future revival.
ASC (aldehyde-stabilized cryopreservation) – chemical preservation of the brain (Nectome's method); performed on a living, anaesthetised person, and therefore fatal.
Gradual replacement (Moravec) – the hypothetical replacement of the brain with artificial components piece by piece, with no moment of "death".
Teletransportation paradox – a thought experiment (Lem, Parfit): a perfect copy of a person is convinced it's the original, while the original dies anyway.
Hard problem of consciousness – the question of why information processing is accompanied by subjective experience; unsolved.
Digital twin – a chatbot trained on a person's data; it imitates style but contains no consciousness or memories.

Sources

Science: brain maps and emulation

Companies, money, controversies

Philosophy and law

Film

Transcendence (2014) (Wikipedia)

Fair warning: the answer is neither "yes" nor "no". It's more interesting than you can imagine at this point.

The version for busy people

If you don't have time, here are the key takeaways:

No, today you can't move a human mind into a computer, and nobody is even close. Humanity's best result is a working digital fruit fly (about 140,000 neurons). You have 86 billion. That's a difference of roughly 600,000 times.
The biggest problem isn't computing power. There are three problems: we don't know exactly what needs to be copied to preserve a mind; we can't scan it in less than thousands of years; and even if we pulled it off, the result would be a copy, while you would die anyway.
Myth: "we're already implanting chips that replace part of the brain". Not true. The best implants read signals from around a thousand electrodes and replace no tissue at all. Reading and replacing are two completely different things.
The companies exist, from serious ones (Eon Systems, Neuralink, Merge Labs) to controversial ones (Nectome) to plain hoaxes (BrainBridge). None of them sells immortality. The ones that tried ended in scandal.
The most realistic path to a longer life is biology, not copying your brain onto a hard drive. And even that promises living "longer", not "forever".

If you want to know why, read on.

1. Let's start with the film, because that's where a very bold vision is laid out

2. Three different things we throw into one bag labelled "immortality"

Before we go further, it's worth sorting out the terms, because the media (and company marketing) mix up several completely different ideas.

This piece is mainly about idea one, because that's what stirs up the strongest emotions when scientists talk about "eternal life in a computer".

3. Brain basics: what our "self" is made of

To understand why copying a brain is so hard, you need to know what exactly we'd be copying. As a layman myself, I promise to explain things in the simplest, most understandable way.

synaptic strength (one connection can be barely a whisper, a gentle suggestion, while another is a shout, an overriding compulsion that steers us, and this changes from minute to minute as you learn),
chemistry: neurotransmitters and hormones that globally "retune" entire regions of the brain (which is why you're a different person after coffee, a different one after a sleepless night, and behave differently yet again when you've fallen in love),
glial cells, which for decades were considered the "glue" holding neurons in place, and which we now know take part in information processing; in the examined fragment of the human cortex there were twice as many of them as neurons,
computation inside a single cell: an individual neuron isn't a simple switch, its "branches" (dendrites) perform local computations of their own,
dynamic state: what's happening in the brain right now, not just its static structure.

4. How much data is that? The numbers that frame these hypothetical musings

This is where it gets concrete, and this is where most of the optimism dies.

One cubic millimetre of human brain

The whole brain

To be fair: those are raw images. A finished, processed map would be much smaller. But to have it, you first have to take those pictures.

How these pictures are taken (and why it can't be done to a living person)

The only method that gives synapse-level resolution is serial electron microscopy. It goes like this:

The brain has to be chemically fixed (to put it bluntly, embalmed, before the tissue starts to decompose).
It's cut into slices about 30 nanometres thick. A human hair is about 70,000 nanometres across, so we're talking about slices more than two thousand times thinner than a hair.
Each slice is photographed with an electron microscope.
Billions of images are stitched together into a three-dimensional block, and artificial intelligence is told to trace every neuron through it.
The AI's mistakes are corrected by hand. This stage, so-called proofreading, is estimated to account for over 95% of the total cost. Checking the fruit fly brain map took people a combined 33 years of work (the sum of the working hours of all team members).

How long it would take

5. The lesson of the worm: we've had the full map since 1986 and it still doesn't work

In my opinion this is the most important part of the whole piece, because it shows that the problem doesn't lie (only) in computing power.

The conclusion is brutal and simple. If we can't get 302 neurons running, neurons we've had the map of since 1986, then every claim about running 86 billion is marketing, not science.

And now, to be fair, the other side of the coin, because in 2024 something happened that gives hope and slightly calls the arguments above into question.

6. The breakthrough: a fruit fly that "works" in a computer (2024–2026)

Step 1: the map (October 2024)

Step 2: a model that predicts behaviour (2024)

Step 3: the body (7 March 2026)

What this really means

This is real science and a real milestone. But before anyone pops the champagne, a few caveats.

What about mammals?

7. Who's working on this (and for how much)

World A: brain emulation

World B: brain mapping

World C: brain preservation and cryonics

This is where the part you need to be careful with begins.

World D: brain–computer interfaces (BCI)

This is the only one of these worlds where real products, patients and real money exist.

Remember one sentence about this whole world: these devices read signals from a small number of neurons or stimulate them. None of them replaces brain tissue. I'll expand on that in section 8.

World E: computers made of living neurons

It's fascinating, but it has nothing to do with transferring a mind. It's growing anonymous neurons for computation. Nobody claims that "someone is in there" in those dishes.

Big projects that have ended

It's worth knowing that the two biggest brain simulation programmes in history have already shut down.

Hoaxes and ghost projects

Digital twins

Everything in one table

Who	What they actually do	Money	Stage	How far from "Transcendence"
Eon Systems	Connectome emulation (fly, then mouse)	approx. $3.5M (unofficial)	Insect demo (March 2026)	Enormously far: 600,000 times fewer neurons than a human
FlyWire / MICrONS / H01	Brain maps (fly, 1 mm³ of mouse, 1 mm³ of human)	Public grants, Google	Science	Mapping only; the human sample is 0.00007% of the brain
E11 Bio	Cheaper mapping methods	Philanthropy	Research	Tools, not a mind
Nectome	Chemical brain preservation	Over $200K in deposits, a grant	Research	No readout method; the procedure is fatal
Alcor / CI / Tomorrow Bio	Post-mortem cryonics	Fees; Tomorrow: €5M	Service	A bet on the future; 0 revivals in history
Neuralink	Signal-reading implant (1,024 electrodes)	Over $1.3B	Clinical trials	Reading is not uploading
Merge Labs	Non-invasive interface (ultrasound)	approx. $252M	Lab	"Decades, not years" (their own words)
Cortical Labs / FinalSpark	Computing on living neurons	Product from $35K	On sale	Not applicable: nobody is being transferred
2045 Initiative	The "Avatar" vision	Private	Dead	Didn't meet a single deadline
BrainBridge	CGI video	0	Fake	Not a company
HereAfter / StoryFile / Eternos	Chatbot imitating a person	Varies	Product	It's an imitation, not you

8. The myth of chips that "replace part of the brain"

Why this matters for the immortality question

9. What about a head transplant?

If you can't move a mind into a computer, maybe you can move the brain into a new, younger body?

10. The biggest problem: is the copy still you?

The question: are you in there?

There are a few ways to respond to this:

"Identity is psychological continuity, not specific matter" (Parfit). If the copy has your memories and personality, then it is you in the only sense that matters. Many transhumanists accept this answer. But it's a philosophical decision, not a scientific discovery, and it still doesn't solve the problem of two copies existing at once.
"Let's do it gradually" (Moravec). If we replace the brain piece by piece, there's never a moment of death and a jump. It's the only path that seems to preserve continuity. Except that, as I wrote in section 8, technologically it's the furthest away of all, and still nobody can prove that continuity of function equals continuity of the subject.
"We don't know, and there's no way to check." In my opinion, that's the only honest answer for now.

Does the copy feel anything at all?

The consequence is unsettling: an emulation could be a so-called philosophical zombie. It behaves like you, talks like you, claims to feel, and yet "inside" there's nobody home.

How companies answer this problem

11. Law and ethics: to "upload yourself somewhere", you first have to die

Let's assume you're ready to accept that a copy will wake up in the computer. You still have a problem: the law.

12. Who pays for the "immortality startup"

Following the money tells you more about this industry than following the publications.

If someone wanted to build a company on this

I've thought this through, because the question "what if I gave it a go" comes naturally to anyone who likes building things. The honest picture looks like this.

On the other hand, there are real, honest businesses around this topic where programming and product skills count, not a lab coat:

Tools for connectomics. A reminder: manually correcting the maps is over 95% of the cost. Any AI tool that cuts that cost has customers in labs around the world. It's a market of small but very motivated buyers. (Fun fact: you can join the research yourself by helping correct the fly map through FlyWire; it works like a game.)
The software layer for brain–computer interfaces. Neuralink and its competitors build the hardware. Apps, patient interfaces, decoders and tools for clinics are a separate, growing niche.
Honest "digital legacy". Products for recording your story, voice and memories for your family are fine, as long as nobody claims it's you living on.

13. Film versus reality, point by point

Scene from "Transcendence"	Reality
Electrodes on living Will's head read his mind	Fiction. Electrodes (even implanted ones) read thousands of neurons out of 86 billion. The only method with sufficient resolution requires embalming the brain and slicing it up.
Will's mind "uploaded" into an existing AI (PINN)	Fiction. There's no way to "merge" brain data with a neural network. Real emulation is a separate simulation, neuron by neuron.
Will in the computer is still Will	Unresolved and probably false. It would be a copy; the original dies. The film itself plays on this uncertainty, too.
Digital Will becomes superhumanly intelligent in weeks	Fiction. A brain emulation, if it worked, would be roughly as smart as the original. Speeding it up is a separate, speculative topic.
Nanobots, taking over the network, regenerating bodies	Fiction, all of it.
Reading brain activity with electrodes	Real, but narrow: a cursor, speech, a robotic arm for paralysed people.
Brain map and simulation	Real at the level of a fly (2024–2026) and a grain of sand from a mouse brain.
"Merging" humans with AI	A real research direction (Merge Labs, Neuralink), but as an interface for control, not as a transfer of a person.

The biggest irony: the film shows uploading as a recording of a living person. The only currently realistic route is the exact opposite: death first, then slicing, then the brain map.

14. The verdict: when?

Let's pull it all together. Here are the obstacles, from the biggest down:

We don't know what to copy. The lesson of the worm. Without understanding how the connections work, not just where they are, the map is like a musical score when you can't read notes. It's a knowledge barrier, not a technical one.
The problem of the copy and of consciousness. Even a working upload produces a copy. And whether the copy feels anything, we have no way of checking.
Scanning. Thousands of years on today's hardware, and no non-destructive method.
Data and computation. Hundreds of exabytes of raw images; computing power unknown to within ten orders of magnitude.
Law and ethics. You have to die; a digital person has no rights.
Money and reputation. Funded by billionaires and grants; the word "immortality" often scares serious people off.

And here are the scenarios, labelled as fact, speculation or marketing:

Scenario	Status in 2026	Realistic horizon	Label
Destructive scan + human emulation, "it's still you"	Works for a fly. For a human: no scanner, no knowledge, the copy problem	If at all, many decades; the "it's still you" version perhaps never	Speculation (technology), marketing ("it's you")
Uploading without destroying the brain, while alive	No technological path exists	Science fiction	Fiction
Gradual replacement of the brain with artificial components	We can read from ~1,000 electrodes; we'd need millions of times more, plus an understanding of the neural code	Decades and more	Speculation
Head / brain transplant	Spinal cord doesn't heal, immune rejection, no living patient in history	No closer than uploading	Speculation, partly hoaxes
Digital twin (chatbot)	Works today	Already here	Fact (as a keepsake), marketing (as "life after death")
Cryonics / chemical preservation	Preserves tissue; nobody has ever revived or read anyone	A bet on an unknown future	Fact (preservation), speculation (revival)
Biological life extension	Billions of dollars, real research, first therapies in trials	Years and decades for "longer", not "forever"	Fact (research), speculation (size of the effect)

15. What this means for you

A few practical takeaways to finish with.

A digital twin is not you. If you want to leave your family recordings, stories and your voice, that's a lovely idea. Just don't let anyone convince you it's a form of life after death.

Glossary

Neuron – a nerve cell; the basic processing unit of the brain. You have about 86 billion of them.
Synapse – a connection between neurons through which a signal passes. The brain has hundreds of trillions of them.
Connectome – a complete map of the connections between neurons. We've had one for the worm since 1986, for the fly since 2024; for humans we have 0.00007%.
Whole brain emulation (WBE, upload) – recreating a specific brain in a computer and running it.
Proofreading – manually correcting the errors AI makes when reconstructing a brain map; over 95% of the cost of connectomics.
Serial electron microscopy – the only method that can see synapses; it requires fixing the brain and cutting it into slices about 30 nanometres thick.
BCI (brain–computer interface) – a device that reads or stimulates brain activity, e.g. Neuralink. It reads; it doesn't replace.
Cryonics – storing bodies or brains in liquid nitrogen after death in the hope of future revival.
ASC (aldehyde-stabilized cryopreservation) – chemical preservation of the brain (Nectome's method); performed on a living, anaesthetised person, and therefore fatal.
Gradual replacement (Moravec) – the hypothetical replacement of the brain with artificial components piece by piece, with no moment of "death".
Teletransportation paradox – a thought experiment (Lem, Parfit): a perfect copy of a person is convinced it's the original, while the original dies anyway.
Hard problem of consciousness – the question of why information processing is accompanied by subjective experience; unsolved.
Digital twin – a chatbot trained on a person's data; it imitates style but contains no consciousness or memories.

Sources

Science: brain maps and emulation

Companies, money, controversies

Philosophy and law

Film

Transcendence (2014) (Wikipedia)