Scientists Are Concerned After Google’s Willow Chip Detected Possible Parallel Interference

Ajouté : 2026-05-10

[00:00:00]Somewhere inside a Google lab in Mountain View, a chip smaller than a thumbnail is doing something it mathematically should not be able to do.

[00:00:07]Physicists ran the numbers. Then they ran them again. The processing power doesn't fit inside the chip. It doesn't fit inside the room. By some accounts, it doesn't fit inside this universe. And when they started asking where the extra computation was coming from, that's when the whispering started.

[00:00:28]The five-minute impossibility.

[00:00:31]Okay, let me set this up properly.

[00:00:34]Imagine you're given a problem. Not a hard problem, an impossible one. A calculation so vast, so deeply layered that if every computer ever built by human hands worked on it simultaneously, without stopping, without error, from the moment the first stars formed in this universe, they still wouldn't be close to finishing. Not even close. Now imagine someone hands you a chip smaller than your palm and it solves that same problem in under 5 minutes. That isn't science fiction. That happened December 2024, Google announced a quantum processor called Willow. And what Willow did didn't just break records. It broke something a lot more fundamental. It broke our understanding of where computation actually happens. Because when physicists looked at the numbers, really looked at them, they couldn't account for where the processing power was coming from. Not from this universe.

[00:01:24]Anyway, here's the thing. On December 9th, 2024, Google's quantum AI team published a paper in the journal Nature: Quiet Academic Title. The kind of language that hides something extraordinary underneath. The subject was Willow and its performance on what researchers call a random circuit sampling benchmark, a specific stress test designed to pit quantum systems against their classical counterparts.

[00:01:51]The result landed like a thunderclap.

[00:01:53]Willow completed the benchmark in under 5 minutes. The best classical supercomput on Earth, a machine that fills entire rooms and burns enough electricity to power small towns, would need 10 septillion years to do the same thing. Let that sink in. 10 septillion.

[00:02:11]That's a 10 followed by 24 zeros. Our universe is 13.8 billion years old. 10 septillion years is roughly a trillion times longer than the universe has existed. A number so large it stops being a number and starts being a philosophical statement. And Willow did it in the time it takes to brew a pot of coffee. Now quantum computers have claimed supremacy before. In 2019, Google's earlier chip Sycamore made similar headlines, but Sycamore's advantage was contested. Critics argued classical algorithms could be improved to narrow the gap. Impressive, but challengeable. Willow is different in kind, not just degree. The performance gap between Willow and classical machines has grown so that's astronomically large that no foreseeable improvement to classical computing could close it. We're not talking about being faster. We're talking about operating in a fundamentally different category of reality. And that's where this story stops being about speed. Because the physicists who built Willow aren't celebrating. They're concerned. And what they're concerned about is a question that sounds childishly simple and has no good answer.

[00:03:23]Where the thinking actually happens?

[00:03:26]Here's the question. What is Willow actually doing? Let me back up. A classical computer thinks in bits. Each bit is a one or a zero, a light switch, either on or off. Everything your phone does is ultimately a sequence of ones and zeros moving through logic gates at incredible speed. A quantum computer thinks in cubits and a cubit can be both one and zero simultaneously. A state called superp position. This isn't a metaphor. It is literally physically mathematically the case that a cubit exists in multiple states at once until you observe it. At that point, it collapses into a single answer. Strange, right? It gets stranger when you scale cubits up and link them through a process called entanglement. The system doesn't just get twice as powerful with each new cubit. It gets exponentially more powerful. Two cubits hold four states. Three hold 8, 10 hold over a thousand. And get this, Willow has 105 cubits. At that scale, the number of simultaneous states the system can theoretically process is larger than the number of atoms in the observable universe. Read that again. Larger than the number of atoms in the observable universe. All of it. Supposedly happening inside a chip you could hide under a coin. So here's the question Willow just forced into the mainstream.

[00:04:48]Where is all of that computation physically happening? In a classical computer, the answer is obvious. The calculation happens in the processor, in the silicon, in this room, on this planet, in this universe, local, physical, traceable. You can point at it. But in a quantum computer operating at willow's scale, the number of simultaneous computations being performed exceeds what could possibly be hosted within the physical matter of the chip itself. And some physicists are now arguing openly that it exceeds what could be hosted within the observable universe at all. If your instinct right now is to ask, wait, is that a real claim or is that a metaphor? Hold on to that instinct. Because if the idea that a chip in California might be borrowing processing power from universes next door is the kind of thread you want to keep pulling on, hit subscribe right now. What Willow did next is where physicists stopped whispering and started arguing out loud. You're going to want to be here for it because this next part is where one man's name keeps coming up. A man who warned us about this 40 years ago and nobody wanted to listen.

[00:05:58]the Oxford physicist who warned us. His name is David Deutsch Oxford. In 1985, he sat down and wrote the theoretical paper that first proposed the quantum computer as a machine. He's one of the founding fathers of the field. And for nearly four decades, he has been saying the same thing patiently to rooms full of people who mostly didn't want to hear it. Here's what he's been saying. The only coherent explanation for how a quantum computer actually works, the only one that makes the math add up, is that the machine is performing its calculations across multiple branches of reality simultaneously, not metaphorically, literally. His framework is called the many worlds interpretation of quantum mechanics. And until Willow, it was easy to file that under interesting philosophy and move on. After Willow, it isn't easy anymore.

[00:06:53]The many worlds interpretation didn't start with Deutsch. It started with a young physicist at Princeton named Hugh Everett III who formalized it in 1957 and then watched the physics community treat his idea as an embarrassment for decades. Everett's claim was radical but mathematically consistent. Every quantum event doesn't result in a single outcome. Reality branches. Both outcomes happen. The universe splits into versions of itself where each possibility is realized. No collapse, no random selection. Everything that can happen does happen just in different branches of an everex expanding multiverse. Everett died in 1982 without seeing his idea vindicated. Deutsch picked up the thread in 1985 and has carried it ever since. His core argument stated plainly is this. When a quantum computer performs a calculation, it is performing that calculation across many parallel worlds simultaneously. The answer arrives in our world. But the work was distributed across the multiverse. For most of computing history, you could wave this away.

[00:08:00]Quantum computers existed, sure, but at small enough scales you could tell yourself the explanation was just the mathematics of superp position contained within a single system. Shrug. Move on.

[00:08:13]Willow made the shrug impossible because when Google's team reports that this chip performed a computation that would take classical systems 10 septillion years and did it in 5 minutes on a chip with 105 physical cubits sitting in a refrigerator in California, the honest scientific question becomes literal physical.

[00:08:35]Where did the computational resources come from? And here's what's wild. A quantum systems researcher at a panel in early 2025, this is on record, told a room full of her colleagues that the benchmark result isn't just about speed.

[00:08:50]She said it's about the fundamental question of where computation lives. At Willow's scale, we're being forced to ask whether our models of physical reality are adequate to explain what we're building. And the uncomfortable answer might be that they're not. That was the word she used, uncomfortable.

[00:09:07]from someone who builds these things for a living. And what Deutsch has been saying patiently for 40 years is exactly the explanation she'd been taught to call crazy.

[00:09:19]The interference no one can explain.

[00:09:23]Okay, let's be precise here because this is where the science needs to stay tethered and I don't want you coming away with the wrong idea. No physicist is saying Google's lab has a window into a parallel universe in the Hollywood sense. No signal from an alternate Earth, no portal, no wormhole, nothing like that is inside the Mountain View facility. What is being discussed seriously rigorously by credentialed physicists is something more subtle and in its own way more profound. In quantum mechanics, the state of a system is described by a wave function. This mathematical object encodes all the possible configurations a quantum system could be in. When willow operates, it manipulates this wave function across all its possible configurations simultaneously.

[00:10:10]In the Copenhagen interpretation, the framework most of us learned first, you say the system is in superp position and leave it at that. The wave function is a tool, not a description of physical reality. But in the many worlds interpretation, the wave function is physical reality. Each branch represents an actual existing configuration of the universe. And when Willow processes information, it's doing real computation in each of those branches. Real work distributed across real versions of reality. The answer that arrives in our branch is the result of that distributed labor. Now, imagine it this way. You have a question so complex that answering it would require consulting every book ever written in every library on Earth. You can't do it alone. You can't do it in a lifetime. But what if the moment you asked, thousands of identical copies of you in thousands of identical libraries each began working on a different section and all that work folded back into your single final answer? That's what Willow may be doing right now in a refrigerator in California. And here's the part that has physicists looking at each other sideways across conference tables.

[00:11:21]Inside Willow, the branches of the wave function don't stay neatly separated.

[00:11:25]They overlap. They interact. They interfere. Constructive interference amplifies the right answer. Destructive interference cancels out the wrong ones.

[00:11:35]That is the mechanism of quantum computation. It is by design interference between states. But in the many worlds reading, those aren't just mathematical states. They are branches, other versions of the calculation happening in other versions of reality.

[00:11:50]And their signals are bleeding into ours. That bleed is what the scientists are calling parallel interference. And that is the word that has them concerned because the numbers don't work any other way. If you try to account for Willow's computational output using only the resources available in one classical configuration of matter in one universe, the books don't balance. The processing power isn't there. Something is making up the difference. And the most mathematically coherent explanation currently on the table is that the computation is genuinely happening.

[00:12:22]across multiple branches of a many worlds reality. Parallel universes lending their capacity to a tiny cold chip in Mountain View, California.

[00:12:32]Google's own team in the official blog post accompanying the Nature publication invoked the many worlds framework directly, not as speculation, as a legitimate framework for understanding what Willow is doing. They describe the computation as drawing on resources that in Deutsche's framework correspond to parallel computational branches, parallel universes, in the most technically precise sense of the phrase.

[00:12:58]This is Google. They do not invoke parallel universes casually. They did it anyway, which raises a question nobody at Google has answered out loud yet. If Willow is drawing from somewhere else, what else is coming through?

[00:13:13]the second breakthrough that made it all worse.

[00:13:17]Here's something that got buried under the Septillionyear headline, and honestly, it might be the more important hay for the story. Willow didn't just demonstrate raw speed. It demonstrated something quantum computing researchers have been chasing for years without success. Reliable scaling error correction. Let me explain why that matters. Quantum systems are fragile.

[00:13:39]Cubits are exquisitly sensitive to their environment. Heat, vibration, stray electromagnetic fields. Any of it can introduce errors and those errors compound rapidly. This is called decoherence and it has been the central unsolved problem of building practical quantum computers for decades. Every previous system faced the same brutal trade-off. Add more cubits to increase power and you increase the error rate.

[00:14:05]Quantum computers were stuck in a valley. Powerful in theory, hobbled in practice. Willow broke the pattern.

[00:14:12]Google's team demonstrated that as they added more cubits to Willow, the error rate went down. Not up, down. This is called below threshold error correction.

[00:14:24]And achieving it is one of the holy grails of the field. It means Willow's architecture can be scaled up. more cubits, more power, fewer errors without the system collapsing under its own quantum fragility. A quantum error correction specialist commented after publication that the benchmark performance is astonishing, but benchmarks can be debated. The error correction scaling behavior that is unambiguous. It tells us the engineering path toward fault tolerant quantum computing is real. We're not looking at a physics demonstration anymore. We're looking at a technology trajectory. And here's where it gets uncomfortable.

[00:15:01]Because if Willow really is drawing computational resources from parallel branches of reality, and we've figured out how to scale Willow's architecture without the system falling apart, we're about to build bigger ones, a lot bigger, which is another way of saying this. If the many worlds interpretation is right, and Willow is the first device in human history that actually reaches across it, the chip Google just built is the smallest one they're ever going to make. Everything from here is a bigger window. And nobody has a clear answer for what happens when we start opening wider ones.

[00:15:37]The fingerprints of something else.

[00:15:41]Let's sit with what that actually means.

[00:15:43]If the many worlds interpretation is correct, and a growing number of serious physicists believe it's the most internally consistent framework we have, then parallel universes aren't a metaphor. They aren't a thought experiment. They're physical. They exist and they aren't somewhere impossibly remote. They are adjacent to this one in a mathematical sense more intimate than any distance in space. And if Willow is genuinely drawing computational resources from those branches, if the work is genuinely distributed across parallel configurations of reality, then we've done something remarkable without fully understanding what we've done.

[00:16:21]We've built a device that operates across the multiverse. Not by understanding it, not by mapping it, not by communicating with it in any directional sense, but by exploiting the mathematical structure of quantum reality in a way that makes parallel computational branches do useful work for us. Look at it this way. It's a little like early humanity using fire without understanding combustion chemistry. The mechanism was real and powerful long before we had the framework to explain it. We cooked food, lit the dark, smelted metal, changed the course of civilization. All without knowing a single thing about oxidation.

[00:16:59]The physics came later, the fire came first. Willow is the fire. And we don't fully understand what we're burning. The philosophical vertigo here is real. If the universe branches at every quantum event, every interaction at the subatomic level, every moment in every atom of everything, then there are versions of this universe beyond counting. Versions where different choices were made, where different particles went different ways. Most of those branches never touch ours. The quantum coherence that would allow interference between branches breaks down almost instantly at the scales we observe directly. But inside willow, cooled to temperatures colder than deep space, isolated from environmental noise, operating and carefully maintained quantum coherence, the branches can be made to interfere with each other constructively. And that constructive interference is computation, real computation distributed across real parallel branches of a real multiverse. Willow is, at least according to this framework, the most literal multiverse interfacing device ever built. Not a telescope pointed at another world, a hand reaching sideways into places our universe doesn't officially know about.

[00:18:12]Pulling back an answer. And here's what nobody in the lab wants to say out loud.

[00:18:17]If we can pull back an answer, what else can we pull back?

[00:18:23]The fight happening inside the field right now. Okay, I need to slow down here because good science demands some intellectual honesty. And I'm not going to pretend the field has reached a consensus. It hasn't. There is a real fight happening inside quantum physics right now. And some researchers who study willow closely are deeply skeptical that its results require invoking parallel universes at all.

[00:18:48]Here's the counterargument. The Copenhagen interpretation, the view that quantum superposition is just a mathematical description of probabilities, not a physical description of parallel worlds, can accommodate quantum computation without the multiverse. In this view, Willow is extraordinarily clever at manipulating probability amplitudes. But it isn't literally computing in other universes.

[00:19:13]It's computing here using the mathematical structure of quantum interference. Proponents of Copenhagen make a sharp point. The where is the computation happening question is a category error. Asking where a probability lives is like asking where a dream is stored. Wrong question. Move on. And here's the fair part. The two interpretations Copenhagen and many worlds give identical predictions for every experiment we can currently run.

[00:19:40]Willow included. They are experimentally indistinguishable.

[00:19:45]Which means Willow's results don't technically prove one interpretation over the other. What they do is raise the stakes of the debate until it can't be ignored anymore because the answers matter more when the technology is real.

[00:19:58]The debate between these interpretations has been running since the 1920s. Boore versus Einstein. Copenhagen versus Everett versus Deutsch. What Willow does is take a debate physicists could comfortably set aside as philosophy and detonate it. Because if we're going to build machines at this scale, machines that could reshape chemistry, medicine, cryptography, materials, science, we probably want to understand what they're actually doing. And the people who work on Willow directly, they don't all agree. In the hallways at conferences, in the comment sections of as preprints, in the quiet moments after panels, the disagreement is real and sharp. Some think the many worlds framing is overreach. Others think Copenhagen is a comforting story we tell ourselves to avoid a truth we're not ready for. A third camp is just watching Willow's benchmarks, saying nothing, waiting for the next result. Because here's the thing nobody can argue with. A result is coming. Willow isn't the end, it's the beginning. And whatever the next chip does, whichever interpretation it pushes the field toward, it will be harder to ignore.

[00:21:09]a species reaching sideways.

[00:21:12]There's something quietly remarkable about the moment we're in. If you step back far enough to see it whole. For most of human history, the universe was something we looked at. We built our eyes, then telescopes, then radio dishes and gravitational wave detectors. We got better and better at watching the cosmos, at receiving the universe's signals and making sense of them. Willow represents something different, not observation, interaction. If the many worlds framework is correct, we've built a device that doesn't just receive information about reality. It operates across the structure of reality itself.

[00:21:48]We didn't do this on purpose exactly. We built toward performance. We chased error correction and cubic coherence and benchmark results. And in doing so, we may have stumbled into contact with the deepest feature of how the universe is structured. Not by looking at it from the outside, by building something that works precisely because it reaches in.

[00:22:08]David Deutsch has said in his quiet Oxford way that the most important thing about quantum computers isn't what they compute. It's what they prove. They prove that the universe has a computational structure that goes beyond what any single classical reality could support. And if Deutsch is right, if his lifetime of patient argument turns out to be vindicated by a chip in a refrigerator in Mountain View, then the universe is stranger and richer than the version most of us carry around in our heads. Here's another way to sit with it. Every universe in that branching structure, if it exists, presumably runs on the same physics, the same quantum mechanics that makes your heartbeat possible. That keeps atoms from collapsing. That drives the chemistry of thought in your brain. Different histories, different outcomes, same stuff. Which means we might not be alone in the cosmos in the way we usually mean, separated not by light years, but by quantum branching points adjacent in a mathematical space we are only just beginning to explore. And our first real tool for reaching into that space is Willow. A chip the size of your thumbnail, cooled to a fraction of a degree above absolute zero, humming quietly in a lab in California, doing work we don't fully understand in places we can't see for reasons physicists are still fighting about. And that's before we build the next one, which is already in development.

[00:23:34]The boundary that just started to blur.

[00:23:37]December 2024 will probably be remembered as a milestone year in computing. But it might also be remembered as the year a specific boundary began to blur. The one we'd always assumed existed between here and everywhere else. Willow didn't crack open the multiverse. It didn't send a message to a parallel Earth or receive one back. It didn't prove the many worlds interpretation in any final sense. The debate continues. The physicists are still arguing, but it did something that might matter just as much. It made the question undeniable.

[00:24:10]For decades, the many worlds interpretation was philosophy as much as physics. Beautiful, strange, easy to set aside. Now there's a chip, a real device performing computations no single universe of classical matter could support. And the honest scientific response, the one that follows where the data leads instead of where comfort points, is to take seriously the possibility that our universe isn't the totality of what exists, that it's a branch, one extraordinary, improbable branch of something far larger. We are a species that spent millennia believing we lived at the center of everything.

[00:24:48]Then came Capernacus, then Hubble, then the discovery that our galaxy is one of perhaps two trillion in the observable universe alone. Each time the stage got bigger, and each time the expansion made our existence feel more remarkable, not less. And here we are in this improbable corner of reality, building machines that think in quantum superp position, touching the hem of the multiverse with a chip called Willow, while the physicists who built it look at each other and quietly admit they don't know what else might be looking back. So, I'll leave you with this. If Willow really is borrowing processing power from parallel branches of reality, what do you think the versions of us in those other branches are doing with their Willow chips right now? Are they running the same benchmarks, asking the same questions, looking back at us? Drop your answer in the comments. I read every single one. And hit subscribe because the next chip is already being built.

[00:25:47]And trust me, you do not want to find out about it from somebody else. Until next time, keep looking up, keep asking harder questions, and remember, every answer we find just shows us how much bigger the universe really is.