Neuralink Co-Founder: AI research focuses on intelligence, brain-machine research on consciousness. People alive today could live up to 1000 years.

Recently, Y Combinator CEO Gary Tan spoke with Max Hodak, co-founder of Neuralink and founder of Science. In this forward-looking conversation, Hodak detailed the groundbreaking progress Science has made in restoring vision for the blind and discussed how brain-computer interfaces (BCI) could serve as an adjunct narrative for longevity and healthcare, ultimately transforming human consciousness.

Here are the key points summarized by Wall Street Insights:

1. Brain-computer interfaces (BCI) are entering the “takeoff era.” They are not just a single product but a broad category like “pharmaceuticals,” covering fields from vision restoration to digital drugs and consciousness merging.

2. Science’s Prima implant has helped over 40 blind individuals regain coherent visual imagery, marking the first time humans have created formed images in their minds.

3. The brain is a computer wrapped in a skull; AI and neuroscience are converging. The “latent space” within AI models is highly similar to how the brain processes information.

4. “Bio-hybrid” interfaces will establish ultra-high bandwidth connections by growing living neurons within the brain, similar to the “braid” interface in Avatar.

Bringing Sight to the Blind: From Drug Discovery to Neural Engineering

For a long time, biotechnology has advanced incrementally, but Hodak believes we are now in a nonlinear “takeoff era.”

Science’s core product, Prima, has completed large-scale clinical trials, with results published in the New England Journal of Medicine. The technology involves a tiny 2mm x 2mm silicon chip implanted under the retina, paired with glasses equipped with a camera and a laser projector that directly stimulates bipolar cells.

“This is the first time a coherent, formed image has been created in a human’s mind,” Hodak states. Previous approaches (like Second Sight) only produced sporadic flashes (phosphene vision) without forming meaningful images.

Hodak emphasizes the advantages of “neural engineering” over traditional “drug discovery”: traditional gene therapies or drug development often take decades and are prone to failure, whereas neural engineering bypasses cell death causes by directly inputting signals into the brain “computer.”

AI and Neuroscience: The “Latent Space” of the Brain

As a computer scientist turned entrepreneur, Hodak views the brain as a computer outside the von Neumann architecture.

He points out that advances in AI are propelling breakthroughs in neuroscience:

• Latent space consistency: When training AI models, their internal representations are astonishingly similar to how the cerebral cortex (like the inferior temporal cortex) processes objects and faces.

• Brain as an API: All sensory inputs (optic nerves, spinal nerves, etc.) can be viewed as APIs of the brain. The essence of BCI is understanding and utilizing these APIs for bidirectional communication.

“Those who say AI is just ‘random parrots’ have no idea what they’re talking about,” Hodak says, “Neuroscience and AI are undergoing deep integration.”

Next-Generation Frontiers: Bio-Hybrids and “Artificial Nerves”

Looking 5 to 10 years ahead, Hodak envisions a blueprint beyond current electrode stimulation—bio-hybrid neural interfaces.

• Evolutionary engineering: Instead of forcing metal wires into the brain, Science is cultivating living neurons induced from stem cells within implants, allowing them to grow together with existing neurons.

• Low immunogenicity: Genetic engineering makes these foreign neurons invisible to the immune system, enabling universal, non-custom grafts.

• Avatar-like connections: This technology could establish a new “neural nerve”—for example, directly connecting the internet to the brain, enabling deep coupling of human consciousness and machines.

Additionally, Science is working on a project called Vessel, aiming to miniaturize complex extracorporeal life support systems (like ECMO), transforming them from “rescue devices” into portable “end-of-life treatments,” thereby redefining medical boundaries.

Entrepreneurial Reflections: From Software to “Wetware” and Elon Musk’s Legacy

Reflecting on his startup experience with Neuralink, Hodak considers it “the ultimate entrepreneurial PhD.” He discussed his collaboration with Elon Musk: Musk recognized early the threat of AI and believed humans must upgrade via BCI to avoid being left behind in the intelligence race.

For young entrepreneurs, Hodak offers two core pieces of advice:

2. Maintain high agency: After setting goals, seek “backdoors” or unconventional pathways to top labs or projects.
3. Seek “oral tradition”: Work early for visionaries like Musk. In Silicon Valley, excellence and game rules are often passed down informally, which can dramatically shape your career trajectory at a young age.

The Event Horizon of 2035

In the final part of the interview, Hodak makes a startling prediction: “The first people who can live to 1,000 years old may already have been born.”

He believes that by 2035, humanity will reach an “event horizon,” after which AI and BCI will enable us to redefine the human condition.

The proliferation of BCIs will follow a path from “severe disability” to “normal aging” and then to “human enhancement.” When implants can offer enviable capabilities—such as direct internet access or super-sensory experiences—the risk-benefit ratio will fundamentally change.

Future may see “digital sleeping pills” or “digital stimulants” that use ultrasound to stimulate specific brain regions, regulating mental states as alternatives to traditional drugs.

Hodak envisions that by 2035, technology will enter an unpredictable “event horizon,” where interfaces connecting humans and machines will be radically redefined.

The full interview transcript follows:

Gary Tan: Welcome back to a new episode of How to Build the Future. Today, we have a heavyweight guest: Max Hodak, co-founder of Neuralink and founder of Science. Science is one of the most exciting BCI companies we’ve seen. Max, welcome to the show.

Max Hodak: Thanks for having me.

Gary Tan: Recently, Science announced that over 40 people have undergone your initial BCI treatments, restoring their vision. What’s the story behind that? What exactly happened?

Max Hodak: Yes, we completed a large-scale clinical trial last year, with results published in the New England Journal of Medicine. It involves a tiny silicon chip, 2mm x 2mm, implanted under the retina. It’s basically a small solar cell array. Patients wear glasses with a camera, which captures the outside world, and a laser projector that projects images into the eye. The laser hits the implant, and the solar cells absorb the light, stimulating the cells directly above. It’s a retinal stimulator. This allows us to bypass degenerated rod and cone cells—the cells normally sensitive to light—and send visual signals back to the retina, assuming blindness was caused by loss of these cells. We conducted large-scale trials across 17 centers in Europe, with very promising results. We’re now applying for market approval, hoping to get it approved later this year.

Gary Tan: For viewers unfamiliar with “brain-computer interfaces,” what exactly are they? What could they do in the past, and what can they do now?

Max Hodak: The brain is a powerful computer, but it’s wrapped inside a skull and doesn’t magically connect to the outside world. It has only a few pathways connecting it—these pathways give you sensory input and motor control. But you might ask: do we want to replace these functions with other things? For example, applications like “virtual reality” or “The Matrix.” Another is restoring lost functions—this is how they’re used today: if someone is blind, you can restore vision; deafness, restore hearing; paralysis, restore movement. Then there’s the concept of “structural neural engineering,” an area not yet heavily explored: can we add new brain regions? Is there a way to understand what’s happening inside the brain to build smarter machines or develop treatments for depression or addiction?

Gary Tan: I’m moved. Currently, it’s mostly about helping those with disabilities or diseases regain normal abilities. I see parallels in AI: previously, computers lacked pure cognition and neurons; suddenly, with many neurons, we have AGI—restoring capabilities similar to humans. And then, beyond that, there’s ASI (superintelligent AI). Have you thought about what future BCIs might look like?

Max Hodak: There are many types of BCIs. They will truly become a broad category like “pharmaceuticals,” not just a single product. I don’t think there will be a universal BCI implanted in everyone. Different modes will suit different purposes. For example, I don’t work on ultrasound, but I believe in the future, ultrasound could enable “digital sleeping pills” or “digital stimulants.” Can you stimulate specific parts of the brain to induce focus or sleep? If that’s possible, I wouldn’t be surprised. It’s more of a consumer-level application.

Gary Tan: So, hopefully, no craniotomy needed?

Max Hodak: Currently, high-quality ultrasound does require drilling through the skull, but I think that will be overcome. For implantable BCIs, it’s serious brain surgery. Recognizing that is important. So, when considering how to apply it to humans and who will use it, initially it will be for severely disabled patients. You always start with the highest risk-to-benefit ratio, focusing on the most disabled— even basic functions can bring huge benefits. I don’t think you or I would want to undergo major brain surgery now just for a cortical motor decoder, since keyboards and mice work very well, with higher performance. Speech can reach about 40 bits per second; many can type at 20 bits; current cortical motor decoders at 10 bits/sec don’t improve daily life. I wouldn’t accept such invasive surgery for that.

But as it becomes more powerful, and we can bidirectionally access richer representations from more brain regions, the risk-benefit ratio will shift. My view is that healthy 30-year-olds probably won’t get implants soon, but eventually, everyone ages. Aging means everything declines. At a certain age, restoring lost functions makes implanting reasonable. Ultimately, this will cross that threshold—you’ll see people who were unfortunate now gaining abilities you envy. That’s when the change begins.

Gary Tan: Let’s talk about those who have never had sight. Why can’t the optic nerve be connected? Can it be remedied later? What role does neuroplasticity play? Do you need to implant BCIs at a very young age when the brain is still plastic?

Max Hodak: Neuroplasticity is fascinating but often misunderstood. During early development, there are “critical periods”; if missed, some connections become very difficult later. There are cases of congenital blindness, but not because the optic nerve is missing—it’s due to congenital cataracts. Their vision is blurry from birth and never forms images. When cataracts are cured in adulthood, results are poor. Their brains can’t interpret the signals, leaving them overwhelmed. Some have even committed suicide because of this. So, early development has clear critical periods. Still, adult brains are more plastic than most think.

Gary Tan: That’s reassuring.

Max Hodak: Yes. If I placed electrodes anywhere in your brain during surgery and woke you up, showing you a flickering light proportional to a neuron’s firing, within minutes, you could learn to control that neuron. The brain is highly plastic under feedback. Part of how cortical motor decoders work is exactly that: decoding signals representing hand or arm movement, and when giving feedback, the brain adapts. Early experiments used just a few neurons with fixed weights: when a neuron fired more, a dot on the screen moved up; less, moved down. The brain learns on its own. Again, under feedback, the brain’s plasticity is immense.

Gary Tan: That’s a powerful moment. Two learning systems interacting, not just a bunch of fixed “if” statements.

Max Hodak: Exactly. The brain is excellent at extracting meaning from information. One reason it’s thought to be less plastic in adulthood is that it’s well adapted to reality. Imagine it as an energy surface with hills and valleys. During normal development, most people’s brain has a large basin—once you fall into it, your state is stable because you’ve adapted. Even if you watch strange movies, you stay in that basin. Some theories suggest psychedelics “anneal” the brain, shrinking the surface to allow other states, but after the effects wear off, it snaps back into the deep energy well. So, even with plasticity, the brain remains in a stable attractor system, making plasticity less apparent. That’s an evolutionary adaptation.

Gary Tan: It’s definitely a selected trait.

Max Hodak: Yes. There’s a tension here: plasticity must persist, or we couldn’t learn new things; all memories are essentially based on plasticity. We experience dramatic plasticity, but with limits—especially after critical periods, in how different brain modules connect.

Gary Tan: I have thousands of questions. I’m very curious about the subjective experience (“qualia”) of those who have received Prima. What is it like with this bio-hybrid approach? Is it like having a second screen? I’m very curious.

Max Hodak: Regarding Prima’s plasticity: during blindness, the brain craves sight. Everything you experience is a constructed world model—your brain’s generative model of reality. When it doesn’t get input from the optic nerve, it still tries to see, increasing “gain” and producing noise. That’s why blind people often report hallucinations. When you first turn on the implant and shine laser light, they say, “Oh, I see flashes.” You can do an experiment: turn on the laser, they see flashes, and play a tone simultaneously. Repeat, then turn off the laser but keep the tone—they’ll say, “I saw flashes.” So, in the first hours of recovery, they must learn to distinguish real signals from hallucinations because the brain’s gain is too high. Learning to differentiate real input from the optic nerve takes training. Prima’s experience is normal vision, though currently black and white, with a smaller field of view, but it’s vision.

Deeper questions concern what a high-bandwidth bio-hybrid neural interface’s qualia would be like—almost unimaginable. We’ll know once these devices are built. But there are natural cases: in Canada, a pair of conjoined twins share one head with four hemispheres. Interestingly, each twin’s two hemispheres are connected normally, but they’re linked by a “biological cable” connecting their thalami. Through this connection, they can share some aspects of consciousness. A less-studied question is: can they see through each other’s eyes? Does this create a new kind of visual field? How do they experience it? Most people have two modes: visual input with eyes open, and imagination. Do they have three or four modes? Or internal monologue? They seem to have internal dialogue but can also communicate through this channel. They can coordinate actions without speaking, with full awareness.

And they don’t get confused—unlike schizophrenics who mistake internal voices for external. They can distinguish. But they experience it directly. So, the question is: is that cable transmitting information in a traditional way, or is some “phenomenal binding” happening? Like the binding of two hemispheres into one moment. These natural cases show some very interesting possibilities, but it’s hard to imagine what it would feel like.

Gary Tan: Paint us a picture of the blueprint. If all goes well, what will this technology look like in 5 to 10 years?

Max Hodak: I think we could achieve near-native acuity, like 20/20 vision. We haven’t yet, but I see the path—color vision and filling most of the visual field are possible within 10 years. But beyond that, the core motivation behind our company is comparing “pharmaceutical” medicine with “neural engineering” medicine. It’s much broader than our initial retinal prosthesis. We started with the retina because it’s a huge unmet need.

Humans aren’t great at discovering new drugs. Sometimes, we find something great—like GLP-1 weight-loss drugs—but often, after ten years of effort, clinical trials show no effect. Many spend huge resources trying to prevent blindness progression or restore vision, with little success. A gene therapy costing $1 million per person has minimal effect for very few patients. In our trials, we can enable a patient who hasn’t seen a face in ten years to read every letter on an eye chart. The brain is fundamentally the most important organ, and from experience, we’re better at engineering it. So, I believe this can lead to a fundamental overhaul of medicine.

Gary Tan: I recall reading about 10 or even 20 years ago that direct electrical stimulation of the optic nerve was possible but with very low resolution and highly invasive.

Max Hodak: Producing flashes (phosphene vision) is relatively easy. Ten years ago, a company called Second Sight implanted stimulators in the eye. It was a 4.5-hour surgery with a titanium box on the side of the eye. They stimulated the cell layer different from ours. They could make patients see flashes—“there’s a flash here,” “there’s a flash there,” forming a letter “A.” But the brain couldn’t combine these flashes into a complete “gestalt.” Similarly, stimulating the visual cortex at the back of the head can produce flashes, but more like a psychedelic effect, not a formed image. To my knowledge, our clinical trial is the first to create coherent, formed images in the human mind.

Gary Tan: Does age-related macular degeneration have features that make this possible?

Max Hodak: Causes of loss of rods and cones include AMD, retinitis pigmentosa, Stargardt’s disease, etc. Age-related AMD affects over 200 million people worldwide. Our device is somewhat agnostic to the cause of photoreceptor loss. We believe it can be used for other conditions too. We’re about to start new clinical trials for genetic retinal diseases affecting younger people. This ties back to “drug discovery” versus “neural engineering”: if you develop drugs, you must focus on the molecular causes inside cells, which vary by disease. Here, we don’t care why cells die; we only care if we can reintroduce visual signals into the brain “computer.”

Gary Tan: I find it fascinating that, as a computer scientist, you’ve spent so much time thinking about input and signals. It seems this mindset is shifting from software to “wetware” (biological systems).

Max Hodak: Yes, the brain is a computer. Saying that might get me flamed in some circles, but I believe it literally. Its architecture is very different from a von Neumann electronic computer, but it processes information. All input and output to the brain happen through a few nerves. The optic nerve, we can call “Cranial Nerve 2”; auditory and balance nerves are “Cranial Nerve 8”; and 31 pairs of spinal nerves control muscles and sensation. You can think of these as the brain’s API. The brain isn’t magically connected to the environment; it’s just electrical pulses on these nerves. In that sense, you have a well-defined interface.

As for how information is processed after entering, it’s extremely complex. It constructs everything we experience. It’s important to realize: you experience yourself in the world, see walls, rooms, lights, but you’re not directly experiencing them—you’re experiencing a world model generated by the brain. The exciting part about AI progress is that we see a unification of neuroscience and AI. We’ve learned more from AI research than expected. Ten years ago, we thought AI researchers learned from neuroscientists; now, it’s the other way around.

Gary Tan: I’m curious—Second Sight produces flashes. How did you identify this API? If I were reverse engineering, I’d try to measure signals.

Max Hodak: BCI development is limited by your ability to record and stimulate signals. As long as you can record, neuroscience is relatively straightforward. The Second Sight example is very instructive. The retina has three main cell layers: 150 million rods/cones connect to 100 million bipolar cells, which connect to 1.5 million ganglion cells (the optic nerve). We stimulate the middle layer—the 100 million bipolar cells—while Second Sight stimulates the last layer—the 1.5 million ganglion cells. They try to bypass the 100-fold compression. The retina performs extensive computation at the last layer. In the first layer (rods/cones), signals are like a bitmap image—like tiles laid out. But in the last layer, signals are compressed, containing edges, motion, color, etc. Stimulating a single cell there produces not a pixel but a complex gradient feature. Since we don’t know the “decoder” or have precise control, the result is a jumble of flashes. Our empirical research shows that stimulating bipolar cells with images can produce images in the mind because that’s a key processing step in the retina.

Gary Tan: Did you know it would succeed from the start, or did you try different sites?

Max Hodak: When we started, unlike most medical device companies, we didn’t base our approach on a patent or academic result. We held a “neural engineering-centric” view. We believed the most valuable thing we could build soon was a retinal prosthesis. By 2021, we thought the technology was mature enough. We used a “first principles” approach. But in biology, you must be cautious with first principles because they’re incomplete—you need to understand what evolution has done.

We faced a choice: stimulate bipolar cells or the optic nerve? Use electrical stimulation or optogenetics? We explored all four quadrants. We quickly found stimulating the optic nerve was very difficult because it requires calibration of hundreds of thousands of degrees of freedom per patient, which is impractical. So we shifted to bipolar cells. Then, whether to use electrical stimulation or optogenetics— we pursued both. We developed advanced optogenetic gene therapies internally. Optogenetics uses proteins to make neurons respond to light. Traditional proteins need strong lasers, but we found highly sensitive proteins that respond under indoor lighting. But clinical application is still 5-7 years away, with many uncertainties. We also researched the best electrical stimulation tech globally; Stanford invented a method ten years ago, later developed by a small European company, which we acquired a few years ago. This “bird’s-eye” research path brought us here.

Gary Tan: That’s incredible, so cool. I want to go back to how you started in tech. How did you begin? Many might wonder, I’ve heard a lot about B2B SaaS, but how can I become like you?

Max Hodak: I come from a software background; my core skill is software. Though I have a biomedical engineering degree, I’ve been programming since I was young. My parents tell a story: I sat crying on a bookstore floor until they bought me a “Learn Visual Basic” book. I’ve always been interested in the brain, inspired by sci-fi, especially The Matrix. The “bit world” is fascinating because building things in the physical world is hard, resource-limited, and Earth is small. But in machines, there are no such constraints. If you can simulate a world, anything is possible. And if you realize you can’t distinguish reality from simulation, then the only thing that matters is the brain. As long as you can engineer the brain, everything else can be replaced. That’s a profound insight.

Gary Tan: That involves providing the right inputs to consciousness…

Max Hodak: It also involves the question of what consciousness is. Currently, some say BCI is an adjunct to AI, aiming to merge humans and machines. I think that’s valid, but more directly, I see BCI as an “adjunct” to “longevity” and “healthcare.” If AI’s ultimate goal is superintelligent machines, then BCI’s ultimate goal is “consciousness machines.” Perhaps there’s no measurement to tell if something is conscious; the only certainty is that you are conscious yourself. If that’s the case, studying consciousness requires experiencing it firsthand through BCI. Once you understand the physical basis of how the brain supports consciousness, you can eventually create superintelligent conscious machines, connected via ultra-high bandwidth links. This perspective differs significantly from the typical BCI narrative.

Gary Tan: We’re at the beginning of all this. Current bandwidth is still relatively low but will improve. Like the PC revolution—who would have thought it all started with a small blue box?

Max Hodak: Still, a bit of “suspension of disbelief” is needed. Biotechnology has always been incremental. You can spend ten years on very subtle progress. But I believe we’re now in a takeoff era. Remember, this isn’t just starting in 2019 or 1999—it began in the late 1800s with the Industrial Revolution. Before that, life changed little for thousands of years; people had no concept of “progress.” They couldn’t imagine how much life would change in the first 15 years after the steam engine. That’s my outlook for the next 15 years.

Gary Tan: You now have electrical stimulation and “bio-hybrid” approaches. Would you call this V2?

Max Hodak: That’s a completely different field. Electrical stimulation of Prima can’t cure glaucoma because it’s a loss of the optic nerve itself. But we might solve it with “bio-hybrid BCI.” We have three R&D pipelines at Science: first, retinal research (Prima); second, neural interfaces; third, vascular projects (Vessel). The idea behind bio-hybrid neural interfaces is: if your brain is a bunch of neurons, how would nature solve this? Evolution is a much better engineer than us. Intuitively, since the brain has two hemispheres, but you experience a unified moment, that’s thanks to the corpus callosum (about 200 million fibers). I wonder: if nature wanted to build a ultra-high bandwidth “brain-to-brain” connection or a new “brain nerve,” like an “internet nerve,” how would it do it? It would grow a new nerve bundle. Our approach is to cultivate living neurons (induced from stem cells) within implants, letting them grow together with existing neurons, forming new biological connections.

Gary Tan: Are they related to your own neurons?

Max Hodak: That’s a deep research area. We use a universal cell line. The most advanced tech is that we “hide” these cells from the immune system. We’re one of the few companies with “low-immunogenicity stem cells.” You don’t need to customize for each patient—that’s too expensive and slow. We engineer these neurons into devices and “graft” them onto the brain. No need to run wires inside the brain or genetically modify your own brain. Techniques like optogenetics or ultrasound often require gene therapy, which is a “one-way door”—if something goes wrong, it’s serious. Here, the only thing edited are the grafted cells; if they die, you’re not worse off. But this opens the potential for growing and forming extensive biological connections in the brain. We’ve seen this in animal models, though not yet in humans. Have you seen James Cameron’s Avatar? Like the alien braid, I imagine a huge new neural nerve with connectors.

Gary Tan: Like a USB port. During Neuralink, what did you learn?

Max Hodak: In many ways, it’s the ultimate “entrepreneurial PhD.” It’s about executing a highly complex technology company, requiring multidisciplinary teams and infrastructure.

Gary Tan: What was V1 like? What language was the brain speaking?

Max Hodak: From an information processing perspective, the brain is full of “representations.” For example, the representation of hand movement: when you open your fingers, a neuron fires; when you close, another fires. In primary motor cortex (many BCI companies record here), signals correspond to understandable things like hand state. Large language models (LLMs) help us understand: when near input or output (muscles, retina), representations are concrete—color, frequency, torque. But deeper in the brain, representations explode into highly abstract concepts. For example, in the inferior temporal cortex, representations are like a “map” of objects or faces. You can think of it as a long string of numbers, where one point represents “vase,” another “Eiffel Tower.” Moving along this “manifold,” you generate perceptions of any object. That’s the “possible object space” represented by millions of neurons.

Gary Tan: Sounds like latent space.

Max Hodak: Exactly, latent space. So AI and neuroscience are converging. Interestingly, when training AI models, their internal representations look very much like brain representations. This shows AI researchers are on the right track. Those claiming AI is just “parrots” have no idea what they’re talking about. Many neuroscientists have switched to AI because modeling is much easier.

Gary Tan: That’s good news for you, since Science aims to be the brain’s API.

Max Hodak: Absolutely. The neural activity recorded in the brain is another “latent variable.” If you can translate it into another model, you can do amazing things.

Gary Tan: Let’s talk about early movement decoding.

Max Hodak: That’s a classic task, achieved in the late 1990s. Many current BCIs do this because it’s proven feasible. The current challenge is electronics: how to miniaturize devices, reduce power, and lower heat so they can be sewn into the skin. That’s a major contribution of Neuralink. Previously, devices required external head connectors, risking infection if the skin didn’t close. Achieving skin closure requires highly efficient microelectronics, enabled by the “smartphone dividend”—massive investments by Apple and Samsung in electronics, which we leverage.

Gary Tan: For bio-hybrid approaches, what are the challenges?

Max Hodak: Some might think adding biological cells increases complexity. But BCIs are like drugs—they won’t be a single product. Many companies will target different applications; bio-hybrid solutions might only be necessary for the most advanced uses.

Gary Tan: Let’s discuss Science’s third part: Vessel. It sounds disruptive too.

Max Hodak: That’s our smallest project. It’s about “perfusion,” like extracorporeal circulation systems. I read a case ten years ago: a 17-year-old in Boston waiting for a lung transplant, kept alive by ECMO. Later, he developed complications, lost transplant priority, and was removed from the list. Doctors faced ethical dilemmas: he was alive, could play games, do homework, hang out with friends, but if the machine was turned off, he’d die. They said his family benefited from him being alive, but it raised fairness issues because maintaining him cost $500,000 daily.

I thought, this shows a huge gap between technology and economic deployment. I researched and found many doctors called this a “bridge to nowhere,” suggesting families give up. I asked: why not treat it as “end-of-life” therapy instead of “bridge” therapy? Doctors even yelled at me. But I saw a big opportunity for integration. Now, 75% of liver transplants use similar perfusion tech, avoiding emergency surgeries at 3 a.m. But these devices weigh half a ton and require private jets. We wondered: can we miniaturize it so a kidney fits into an airline’s luggage compartment? Can that 17-year-old carry a “portable artificial lung” home? In his case, they eventually stopped changing the oxygenator filters, and he died of a blood clot after a week. That’s unfair. If we can provide vision, hearing, balance, movement, and help him walk into the world, that’s a way to reimagine medicine.

Gary Tan: How did Neuralink start? How did you meet Elon?

Max Hodak: In early 2016, I received an email from Sam (Sam Altman) titled “Crazy question.” He said Elon wanted to start a BCI company and asked who should lead it. I initially recommended a friend from MIT, but an hour later, I thought: wait, I can do that! Sam introduced me to Elon. Elon already had the idea and the name Neuralink. Later that year, our small team met weekly, and the project snowballed into Neuralink. Many initial members were acquaintances from my Duke University lab.

Gary Tan: What was it like discussing connecting computers to the human brain?

Max Hodak: Elon saw the future of AI earlier and more clearly than most. He believed it couldn’t be separated from humans; it must be integrated to upgrade humans, not leave them behind. Looking at Earth’s history, humans have dominated the planet, confining our closest primates in cages. So, more powerful intelligence is very dangerous. That’s a driving force.

Gary Tan: You’re a model of shifting from pure software to hardware and breakthrough research. What advice would you give your 2016 self?

Max Hodak: Two things. First, I did: set clear goals and have high agency. I knew in college I wanted to do BCI, even though my lab rarely took undergrads, I found a way through independent research in chemistry. Second, I didn’t do: after graduation, I founded a cloud lab company, Transcriptic. It was promising, but I was in “hard mode,” a grind from 2012 to 2016. Looking back, I should have started working for visionaries like Elon earlier—that would have greatly expanded my horizons and understanding of “game rules.” Entrepreneurship is often an “oral tradition” passed down in Silicon Valley’s culture, which can dramatically alter your career path at 20.

Gary Tan: Looking ahead 10-20 years, what excites you most?

Max Hodak: I have a “future event horizon” for 2035. I used to pride myself on predicting the future, but beyond 2035, I can’t see clearly. I believe the first people who can live to 1,000 may already be born. Earth is in an extraordinary era of change. AI and BCI are two parallel narratives. People are beginning to realize AI is real but don’t fully grasp its potential; meanwhile, the possibilities of BCI are largely unknown. Overall, I’m optimistic. My “doom probability” (P-doom) is low. By 2035, I can’t say all diseases will be cured, but we’ll have a new way to redefine the human condition, including interfaces connecting humans and machines, and humans and each other.

Gary Tan: Max, thank you so much for joining us and for building the future.

Max Hodak: Thank you, Gary.

Risk Disclaimer and Legal Notice

Market involves risks; invest cautiously. This article does not constitute personal investment advice and does not consider individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions herein are suitable for their circumstances. Invest at your own risk.