Above my pay grade: Jensen Huang and the quantum computing stock market crash
Update (Jan. 13): Readers might enjoy the Bankless Podcast, in which I and Justin Drake of the Ethereum engineering team discuss quantum computing and its impact on cryptocurrency. I learned something interesting from Justin—namely that Satoshi has about $90 billion worth of bitcoin that’s never been touched since the cryptocurrency’s earliest days, much of which (added: the early stuff, the stuff not additionally protected by a hash function) would be stealable by anyone who could break elliptic curve cryptography—for example, by using a scalable quantum computer. At what point in time, if any, would this stash acquire the moral or even legal status of (say) gold doubloons just lying on the bottom of the ocean? Arrr, ’tis avast Hilbert space!
Apparently Jensen Huang, the CEO of NVIDIA, opined on an analyst call this week that quantum computing was plausibly still twenty years away from being practical. As a direct result, a bunch of publicly-traded quantum computing companies (including IonQ, Rigetti, and D-Wave) fell 40% or more in value, and even Google/Alphabet stock fell on the news.
So then friends and family attuned to the financial markets started sending me messages asking for my reaction, as the world’s semi-unwilling Quantum Computing Opiner-in-Chief.
My reaction? Mostly just that it felt really weird for all those billions of dollars to change hands, or evaporate, based on what a microchip CEO offhandedly opined about my tiny little field, while I (like much of that field) could’ve remained entirely oblivious to it, were it not for all of their messages!
But was Jensen Huang right in his time estimate? And, relatedly, what is the “correct” valuation of quantum computing companies? Alas, however much more I know about quantum computing than Jensen Huang does, the knowledge does not enable me to answer to either question.
I can, of course, pontificate about the questions, as I can pontificate about anything.
To start with the question of timelines: yes, there’s a lot still to be done, and twenty years might well be correct. But as I’ve pointed out before, within the past year we’ve seen 2-qubit gates with ~99.9% fidelity, which is very near the threshold for practical fault-tolerance. And of course, Google has now demonstrated fault-tolerance that becomes more and more of a win with increasing code size. So no, I can’t confidently rule out commercially useful quantum simulations within the next decade. Like, it sounds fanciful, but then I remember how fanciful it would’ve seemed in 2012 that we’d have conversational AI by 2022. I was alive in 2012! And speaking of which, if you really believe (as many people now do) AI will match or exceed human capabilities in most fields in the next decade, then that will scramble all the other timelines too. And presumably Jensen Huang understands these points as well as anyone.
Now for the valuation question. On the one hand, Shtetl-Optimized readers will know that there’s been plenty of obfuscation and even outright lying, to journalists, the public, and investors, about what quantum computing will be good for and how soon. To whatever extent the previous valuations were based on that lying, a brutal correction was of course in order, regardless of what triggered it.
On the other hand, I can’t say with certainty that high valuations are wrong! After all, even if there’s only a 10% chance that something will produce $100B in value, that would still justify a $10B valuation. It’s a completely different way of thinking than what we’re used to in academia.
For whatever it’s worth, my own family’s money is just sitting in index funds and CDs. I have no quantum computing investments of any kind. I do sometimes accept consulting fees to talk to quantum computing startups and report back my thoughts. When I do, my highest recommendation is: “these people are smart and honest, everything they say about quantum algorithms is correct insofar as I can judge, and I hope they succeed. I wouldn’t invest my own money, but I’m very happy if you or anyone else does.” Meanwhile, my lowest recommendation is: “these people are hypesters and charlatans, and I hope they fail. But even then, I can’t say with confidence that their valuation won’t temporarily skyrocket, in which case investing in them would presumably have been the right call.”
So basically: it’s good that I became an academic rather than an investor.
Having returned from family vacation, I hope to get back to a more regular blogging schedule … let’s see how it goes!
Follow
Comment #1 January 9th, 2025 at 4:47 pm
This sounds very true for superconducting qubits to me. It feels less true for trapped ion qubits. No idea where photonic qubits stand in this regard.
Which reminds me that I was initially very sceptical about scalable photonic qubits, because they can operate at room temperature. When I read about some nice successes there, I was surprised to learn that they simply went to very low temperatures too. I should have thought of that myself! And I guess the freedom to play with temperature and speed will probably be exploited too for the first really useful applications of quantum information processing.
Comment #2 January 9th, 2025 at 6:52 pm
You once opined you wished you bought some bitcoin. Would Scott Aaronson in future feel similarly about Quantum stocks?
Comment #3 January 9th, 2025 at 7:00 pm
If you believe quantum computing is still far away and will have little practical use, then good, the stocks went down!
If you believe the other end of the spectrum, then good, the stocks went down, and you can buy more!
Comment #4 January 9th, 2025 at 7:01 pm
Bitcoin similarity #2: Very possibly, yes!
When it comes to investing, my problem is that making a mistake and losing my investment would subjectively feel 100x worse to me than getting a 100x return would feel good.
Comment #5 January 9th, 2025 at 11:30 pm
Hi Scott, long time listener first time caller. As a scientist in the field who has also been intimately investigating the quantum investing underworld, I fear that your ambivalent analysis of this situation is too caught up in proof by exhaustion about future events (I.e. the edge case that we have a commercially useful QC in 10 years) that you’re missing the moral imperative of the here and now: quantum computing companies are grossly misrepresenting their technologies at the expense of investors and ultimately the general public.
The QC market didn’t “lose” a few billion from jensen’s comments, that few billion should never have been there in the first place. Every company has played a role in this. Sure Google has made incredible progress with QEC on willow, but making a single logical qubit is not so eye popping to the general public as 25 septililion years vs 5 minutes or parallel universes, is it? Surely the past month of stock explosion owes more to Hartmut’s science-by-blog than the QEC paper that was already on the arxiv in August. IBM in 2023 made that dubious claim of “quantum advantage” with a Time magazine article ready to publish with the announcement, only to be upstaged by several research groups and a guy with a Commodore 64; I would wager those classical reproductions didn’t filter their way to the masses the way the initial announcement did. IonQ right now markets their machines with “algorithmic qubits” and expects AQ64 this year; does anyone other than them know what an algorithmic qubit even is? Their marketing claims that they are “useful qubits you can run algorithms on”, obviously attempting to conjure a relationship to logical qubits; peek under the hood at the non-peer reviewed blog posts and you’ll find that AQ simply means running simplified Hamiltonian simulation/phase estimation circuits and using plurality voting to throw out the bad data, sometimes well over 99% of the total. They have graphs that claim to run 27-qubit QFTs with 100% fidelity (obviously using this post-processing) and marketing will place them on a slide next to similar charts for IBM/quantinuum machines which accurately demonstrate that attempting to run a fault tolerant algorithm without QEC will turn to noise; this is deceptively presented as an apples-to-apples comparison. This is only a selection of the bullshit peddled by companies, not including claims of imminent quantum advantage (running what exactly is never clear) or insisting we already have quantum advantage by running toy optimization problems on a quantum annealer. Whether these claims are driven by ignorance, greed, or some combination, the end result is that the channels by which science reaches the masses are totally flooded by marketing masquerading as science, all with the purpose of extracting funds from investors that couldn’t possibly do reasonable due diligence unless they worked in the field themselves.
I’m a bit miffed by your statement that you could’ve missed this episode of stock market shenanigans had it not been for friends texting you. I feel your comment about being “the world’s semi-unwilling Quantum Computing Opiner-in-Chief” is at odds with your argument with Farhi a few months ago about the burden of being a communicator in the field. Burden goes beyond dealing with a few clueless journalists asking noob questions about the field, the burden is that soulless capitalists have commandeered “your tiny little field” and sold to the public a vision of the future completely incompatible with reality; ultimately someone will pay the price and I doubt it will be them. Your post seems to suggest that even assuming the worst in the industry, the ends may still justify the means. I can’t agree, especially if the means prevent us from getting to the ends or if the ends aren’t good enough. Small instances of Hamiltonian simulation with dubious commercial application might be interesting for scientists (myself included), but if that’s all we have to show for another decade or more of significant public investment was it worth it? To us maybe but to them almost certainly not, unless marketing invents the new use case to string them along. Just ten more years bro I swear. If we meet the Gidney-Ekera resource counts and produce a factoring machine in 20-30 years will that be worth the turbulence along the way? To some. If you lost a job, lost investments because of it, you might think otherwise. We also can’t exclude the possibility that no quantum advantage/utility will be reached.
You’ve serviced the community for a very long time and I think we’re all indebted to this blog in one way or another. Yet, I can’t help but feel as though your perspective is insulated from the magnitude of sludge the average person without a quantum background wades through today. Right now your post is being used by investors as proof-positive that Jensen was wrong and the QC companies have had their timelines right this whole time; I don’t agree and it’s obviously not what you said, but this is the burden.
I hope this post doesn’t read as prescriptive; I do enjoy the schadenfreude when you dismantle outlandish claims from companies and while I think the field would benefit from everyone being less collegial I understand making room for skepticism to give way to optimism on occasion. I am happy with the Google result too. I suppose I just ask that on the days that you are a “willing” quantum computing opiner-in-chief, you are aware that the investing public is force-fed an endless stream of borderline nonsense. The “soul” of the field is no longer nurtured by distinguished academics like yourself, it has been co-opted by predatory technocrats who are selling the public something that doesn’t exist and if we’re not careful may never exist. The rare occasion like yesterday when one of them said something that wasn’t hyperbolic should be celebrated.
Comment #6 January 9th, 2025 at 11:32 pm
Perhaps you knew Al Despain, one of the major early researchers in quantum computing. I knew him from other computer architecture shared interests. One day at a conference we were talking in the hallway outside the auditorium, and I happened to ask him what were the chances that quantum computing would amount to the world-changing revolution that everybody seemed to think it was. His answer was simple and unequivocal: “Three percent.” This was back in the twentieth century. I haven’t seen anything to change my mind since…
Comment #7 January 9th, 2025 at 11:51 pm
I saw an article in some biz mag (maybe Forbes?) today stating that Huang is wrong and millions of $ are being made by using QC right now. it wasn’t clear if this money came from selling quantum computers, or from using them.
Anyone aware of any actual money being made from USING QC?
Comment #8 January 10th, 2025 at 12:38 am
Its clear that Nvidia CEO tells such things because
AI is in some sense competing with Quantum computers:
https://backreaction.blogspot.com/2024/11/ai-could-soon-make-quantum-computing.html
Comment #9 January 10th, 2025 at 2:39 am
The way public statements can so drastically shift valuations in quantum computing really highlights the disconnect between scientific progress and market speculation. Do you think this creates more harm than good for the field overall?
Comment #10 January 10th, 2025 at 4:45 am
This Huang-induced crash came after (and probably in response to) a truly massive increase in quantum computing stock-market valuations that started right after Trump got elected in November. E.g. RIGETTI went up by a factor of nearly 20 in 2 months before the latest crash, D-WAVE by a factor 10, IONQ by a factor >3 etc. Even after the latest crash, the valuations are still *much* higher than they were in early November. What is that all about? Is Trump considered so much better for the prospects of quantum computing than the Democrats?
Comment #11 January 10th, 2025 at 5:52 am
What happened is that the CEO of the largest company specializing in classical computing has just tanked the rival and potentially superior field of quantum computing.
A couple more takes:
Such a huge drop of 40% indicates that the market is as dumb as a school of fish that follow whales in the ocean. Perhaps, they should start thinking how to invest in Greenland instead of QC, it’s the new meta now.
Even 40-qubit quantum computers are useful and practical, but misguided opinions from influencers could easily delay their development by decades. Who cares, though? The development of society is not the end goal of current leaders anyway. It’s all about the redistribution of resources and power.
Comment #12 January 10th, 2025 at 6:53 am
Scott #4,
Most rational people like us would feel the same way, and sit out on things that are out of the ordinary. However, its ironic that humanity’s progress has been dependent on the others, those who take risks that are deemed irrational at that moment, but took that step anyway in spite of the odds stacked against them. Of course close to 100% of those people would have failed and paid the price for it, but the small fraction of 1% that succeeded , made this world what it is today ?
Comment #13 January 10th, 2025 at 8:20 am
+1 that a correction was probably due regardless the cause. I expect 2025 will see large growth in non-sense quantum hyping in connection with non-sense AI hyping so happy to see the year start off with a correction that will drag that trend a bit downwards.
Comment #14 January 10th, 2025 at 8:42 am
Huang is almost certainly correct. It’s interesting that you took Sabine Hossenfelder to task in an earlier post when she made the same basic point though.
Comment #15 January 10th, 2025 at 9:17 am
Dril #5 is entirely right here.
None of these companies should be worth anything and this whole episode has a good chance of destroying the field, which would actually be tragic.
Comment #16 January 10th, 2025 at 10:03 am
Here is IONQ’s statement:
https://finance.yahoo.com/news/ionq-ceo-timeline-quantum-value-124500312.html
Anyone want to estimate the BS / Facts ratio?
Comment #17 January 10th, 2025 at 10:34 am
Time is money.
If you think there is 10% change of $100B valuation in 10 years, the expected value of $10B is not the current value. You must use time value of money to calculate the present value. Assuming typical 8% annual interest. It would be $4.6B in present value , $2B if it takes 20 years.
Another variable is the term of patent. In the US and Europe it’s 20 years from the filing date. Spending R&D capital into something that creates value 20 years from now has reduced value.
Comment #18 January 10th, 2025 at 11:17 am
On BS detection:
I have been actively following science — particularly math, physics, computing, and astronomy; — for over 60 years, both as an observer and as a profession. A disconcerting trend is the increase in bullshit (BS) in conjunction with the widespread lack of intuitive BS detection.
A fond memory is William Beale, who I worked for 50 odd years ago, describing controlled nuclear fusion research as a “classical R&D rathole”, and comparing it with an atomic powered aircraft project he had worked on (hope that’s not a secret, it was a dumb idea anyway).
Despite not having studied QM in over a half century, my internal BS detector has always assigned a low likelihood for “useful” QC results anytime soon (or maybe forever). Some people who know a lot more than me — i.e., dril #5 — have provided some cold hard facts indicating the same.
Comment #19 January 10th, 2025 at 12:22 pm
Scott,
The reality is simple. In mid-December, QC stocks started shooting up for apparently no reason. Jensen’s comment was the catalyst for the air to come out. Not everything that goes down needs a rational explanation if you don’t take into account why it started shooting up. There is a lot of mo-mo trading that goes on.
Comment #20 January 10th, 2025 at 1:09 pm
dril #5 (and others): When I wrote this post, my central goal was to avoid (even indirectly) giving investing advice of any kind: neither “buy,” nor “hold,” nor “sell.” This is because any advice I give might be wrong, and anything I’m wrong about I might get blamed for, and unlike with (say) questions about the fundamental limits of quantum algorithms, here I have no principles to tell me what the right answers are.
As I said, my own money just sits in index funds and CDs. There’s something deep in my emotional makeup that makes me hate betting or guessing, in any situation where I might be wrong in ways that I won’t understand even after the fact. Intellectually, I think Bayesian epistemology has a great deal to be said for it, but I can’t seem to implement it in practice.
And, yes, I do still feel burned by the experience of listening to all the serious, cautious, credible, responsible experts explain the staggering difficulties of AI and the generations-long slog that probably lay ahead—while I discounted the hype-addled crazies who claimed it might be a mere decade or so (!!) before AI would start reshaping the world in the ways foreseen by Turing and Asimov and Arthur C. Clarke … and then seeing the crazies vindicated by events.
How did the crazies see what the serious people missed? In large part, they did it by just naïvely starting at log-log plots, looking for when the flops per second would exceed some crude estimate of what the human brain does (sometime in the 2020s, they said), and ignoring all the sneers about the unsoundness of that methodology.
So then, what happens when I stare at log-log plots for the infidelity of 2-qubit gates? Well, I see a steady decrease over the past 25 years, with the fault-tolerance threshold being barreled through right about now. How many years of additional engineering will be needed after the fault-tolerance threshold is firmly crossed? I don’t know, any more than Bohr and Fermi and Szilard knew the analogous thing in 1940. But just like then, surely the timeline will respond to how urgently people want the thing and how much they’re willing to spend on it.
Having said that, we do need to separate the “timeline” question from the “scope of applications” question—Jensen Huang alluded to both in his comments. To say this as clearly as I can:
If someone invested in quantum computing specifically because they believe it’s poised to revolutionize optimization, AI, finance, and other broad classical domains in the near future—in that case, I strongly advise them to sell. We actually know something about these matters, and what we know directly contradicts what many QC startups have been telling investors and the public.
If, on the other hand, the crux for you is just the number of years until full fault-tolerance, then there’s probably little reason to privilege Jensen Huang’s guess over your own guess, or a thousand other people’s.
Comment #21 January 10th, 2025 at 1:25 pm
Sandro #14:
Huang is almost certainly correct. It’s interesting that you took Sabine Hossenfelder to task in an earlier post when she made the same basic point though.
With Sabine, what irked me was not her estimate of how long QC will take (she’s perfectly entitled to her estimate as you, I, and Jensen Huang are entitled to ours), nor was it anything she said about the applications of QCs (I’m not aware of any disagreements she and I have there). It was purely her remarks that struck me as unduly dismissive of the hardware progress that Google and others have made (“we’re still about a million qubits away from a million qubits”)—as if the whole vast experimental ground covered in the past quarter-century were within a rounding error of zero. That’s as scientifically informative (and generous) as summing up LIGO’s progress in 2014 by saying, “well, so far they’ve detected a grand total of zero gravitational waves.”
Comment #22 January 10th, 2025 at 11:19 pm
Scott #21
You are pretty much an enthusiast and confident that QC will work someday. Maybe it will. I tend to be a pessimist and in this case say “don’t hold your breath”.
Anyway, I regard the analogy with LIGO as pretty weak:
The LIGO plan involved well known stuff that would have to be made much more precise than before, but there was no reason to suspect it would be a dead end. I have been reading about it in, e.g., “Science News Weekly” for decades, and don’t recall any speculation that it would never work.
On the other hand, the difficulty of getting useful QC is more like that of drilling a hole to the center of the earth. Is the vast improvement in the last half century — deeper holes drilled, up to a dozen KM or so by now — within a rounding error of zero? No — but I still doubt if they will ever get there.
Comment #23 January 10th, 2025 at 11:21 pm
1) The Stock Market is very weird in that you are not betting on a company succeeding, our a betting that others thing so, so you can buy before that happens and sell after that happens.
2) There may be some Greater-Fool-Theory stuff going on also.
3) Even if you think Quantum is going to work in (say) a decade and you can wait that long for an investment to pay off, there is still the question of WHICH quantum company is going to have the breakthrough.
4) Is the following legal:
Respected Quantum person buys lots of quantum stock and THEN proclaims loudly
We are 3 years away from quantum being able to do X!
Then the stock goes up, and respected-quantum-person does well.
5) I agree with Scott on risk-aversion (that does not mean either of us is right). As a small
example, if I was on the show Cash Cab and won x dollars then there is NO WAY I
would go for double-or-nothign at the end. The reason extends to more serious investments:
The joy I would feel by getting 2x instead of x is TRIVIAL compared to the sadness I would
feel losing x. This works for ALL x.
Comment #24 January 11th, 2025 at 8:45 am
“what a microchip CEO offhandedly opined about my tiny little field, while I (like much of that field) could’ve remained entirely oblivious to it, were it not for all of their messages”
what a weird way to put it…
“my tiny little field” smells of false modesty (it’s probably sarcasm, I dunno): Scott has been telling us about how much is being invested in QC as a proof that it matters (and if that part of his work didn’t matter enough, he’s been simultaneously working at OpenAI, where Jensen’s chips are kind of a big deal :P).
And the typical ‘subtle’ diss of theorists at the experimentalists and even more at the engineers that make extremely complex things happen.
Jensen Huang is no dummy, he finished high school ahead of time with a strong math background while being an advanced ping-pong athlete, then got his BA in Engineering at 20 and then, while working as a microchip designer during the day, in the evening he got an MS in EE at Stanford. So I’m pretty sure he’s not totally clueless on the topics of QM and QC or whatever else matters (robotics, etc).
Comment #25 January 11th, 2025 at 8:58 am
And of course “classical” computer makers would put a lot of thoughts and have things to say about quantum computers, especially when it’s all chips, using many of the same manufacturing techniques and scaling issue in the end. If QC ever become a ‘product’ (*), it’s not like the current electronic industry is just going to sit on the sideline and what it happen, they’d make sure they’d take a stab at it.
(*) Just like bomb makers in 1945 probably had a lot of thoughts an opinions about the invention of the atomic bomb. It’s all bombs in the end too, but it wasn’t clear whether one was gonna become an actual commodity and overtake the other kind as a practical tool of war.
Comment #26 January 11th, 2025 at 9:11 am
The other thing is that for all the money computer manufacturers (like NVidia) are making, the real value of classical computing has been in the software. There’s a limitless amount of useful programs that can be easily written on the hardware, leading to the wealth boom of the internet, the cellphones, etc. Very quickly the value has been put in everyone’s hands with the revolution of the personal computer in the early 80s. Computing became a commodity accessible to everyone.
With QC it’s not so clear how money will be made, or at the very least it seems it will be more like an old school mainframe, and then various heavy industries will pay a hefty fee to use the compute power to accelerate some very critical but narrow problems (like pharmaceuticals).
Comment #27 January 11th, 2025 at 11:27 am
Scott, I think your sentiments about revolutions in optimization are correct. I think dril is saying the same thing. For example, quadratic speedups (if they happen with NISQ) will not give us a revolution.
I do want to advise a word of caution to the public at large. Personas like dril make an impassioned plee not out of good will but because they themselves hold a financial stake. Getting paid a QC salary while short the market to hedge their career path.
There’s no winning when you’re playing with speculators, long and short, they will work to manipulate you, a figure of authority, into playing to their tune. The only winning move tends to be not to play.
I think the same kind of financial pressure can be said to be a deficit of Jensen. He promotes delusional ideas like their transistor based chips beating moore’s law (hint: they’re not), to keep people confident in their ability to maintain a lead and distract from physics limiting their potential. The era of exponential performance for compute companies is at the tail.
Comment #28 January 11th, 2025 at 1:10 pm
Raoul Ohio #22: A lot of people in the QC startup world would be amused to hear me described as an “enthusiast.” 🙂
But considered as engineering problems, I do think the analogy between LIGO and QC is quite close. In both cases, people understood decades in advance that something was possible, provided the laws of physics continued to work like the textbooks said. And crucially, they could even estimate how many additional orders of magnitude of device precision they would need to make the thing work. The problem was “merely” that the precision was way beyond any technology that then existed. But then people chipped away at it, and closed the gap over several decades.
AI is an important example that wasn’t like this. There, no one had any physical principles by which to estimate the engineering requirements in advance — they could only build things, try them out, and see.
Comment #29 January 11th, 2025 at 7:54 pm
Scott #20: Thanks for responding to my post. I appreciate your perspective on the progress of AI in the past decade as I only started hearing about neural networks in grad school a little over 10 years ago. Do you have any sources/articles of the naysayers from that time? I’d be curious to read their thoughts and see what they missed knowing what we know now.
Surely though the AI situation isn’t totally comparable to what a future QC boom might look like. People have been creating very impressive demonstrations of AI/ML over the past decade, like the google locomotion video, alphaGO, marI/O finding wall warps in sonic, thispersondoesnotexist, etc. etc. Maybe for the general public ChatGPT seems like extraterrestrial technology that fell out of the sky, but for anyone loosely following the situation I feel like there were plenty of signs indicating where we were headed.
In the case of QC, if error rates continue to improve, if we can start building bigger chips at some predictable rate, shouldn’t we expect clues along the way that we’re headed somewhere productive? Right now the QC “crazies” are the ones that are trying to sell the public on the idea that jamming more qubits into bigger and bigger parameterized quantum circuits will give us variational algorithms that outperform classical. We know there’s no theoretical guarantee this will work, there’s plenty of reason to expect that will NEVER work, but if it somehow does pan out (to my chagrin I’ll be honest) surely we’ll see something like a quantum entrant to a SAT competition or a working QML demonstration beyond a toy problem scale in time for you to revise a currently reasonable opinion that QC won’t be useful in the next decade, at least certainly not in the way the “crazies” expect.
This brings me to my other point, you offered caution to investors “in quantum computing specifically because they believe it’s poised to revolutionize optimization, AI, finance, and other broad classical domains in the near future”. Let’s be honest here, that’s basically everyone. No one is invested in the field because they’re excited about academically interesting lattice models; they want to see QC make real dollars for real businesses. Right now companies (both QC and customers) are content to engage in pure PR ventures because they both benefit, not because customers are materially benefitting from running VQE on qiskit. But the public doesn’t appreciate the inner workings of these deals, they just see a headline of a Fortune 50/100 company partnering with QC and think happy thoughts of the future, and industry can keep plodding along selling something “slightly out of reach”.
I guess I’m saying that it is THAT BAD out there. Certain publicly traded companies are selling imminent quantum advantage. They’re not carefully distinguishing between “timeline” and “scope of applications” and engaging in honest discourse with skeptics as you are. If certain QC applications pan out in an optimistic timeline as you hope, it will be in spite of these people and not because of them. I personally wouldn’t give them an inch.
odlird #27: if I was shorting QC stocks I would be having a bad couple of months. I’m with Scott on this, I only hold index funds. current qc investor shills are so wrapped up in the marketing marketing that they’re spinning conspiracy theories about Jensen merely stating what everyone else has been saying for a long time. my “impassioned plea” is primarily driven by a hatred of manipulative bullshit
Comment #30 January 11th, 2025 at 11:23 pm
dril #29
Could you share your opinion on the prospects of QC in the context of quantum chemistry?
Comment #31 January 12th, 2025 at 1:30 pm
dril, aka https://x.com/SPVLABS , you have a history of interacting with financial twitter & other places that implies you may be a little bit disingenuous about your holdings and intentions.
Comment #32 January 12th, 2025 at 6:48 pm
vladimir #30: I am definitely not the person to ask about this. my impression is that you need some way to efficiently encode a chemical hamiltonian (not always the case) and then do something with that hamiltonian, either simulation via Trotterization or QSP or something else or find some way to estimate ground state energy. The Hamiltonian simulation algorithms have something in the way of a provable exponential speedup because you’re basically doing a Taylor approximation to e^ix and every additional degree in your truncated series gives you exponentially better accuracy. there are a shitload of caveats to these sorts of things. offhand estimates I’ve seen of quantifiably financially useful hamiltonian simulation applications have slightly worse resource requirements than Shor, some about at Gidney-Ekera levels and some much worse.
A popular method for NISQ to do eigenvalue estimation is VQE, where you basically take a parameterized quantum circuit, initialize a state close to the ground state, and then classically optimize angles in your circuit to produce a quantum state close to ground. From the outset this method has no provable guarantees of success, and resource estimates I’ve seen of this to achieve quantum advantage even for small chemical reactions are exorbitant. But, it’s pretty easy to implement on a NISQ device and compared to other NISQ applications it “seems” more plausible, at least to non-chemistry person like me.
If I defer to the experts I’ve heard talk about these things, they seem to think Hamiltonian simulation for small molecules is a good potential first use case for quantum advantage. Whether that leads to financial gain is a different story, but I’d by their hunch. What gets presented in the media however is that based on a few molecules of interest (e.x. FeMoco, Lithium?) we can use a quantum computer to solve world hunger, stop climate change (via better batteries), etc. when in reality these problems are very complex politically and a better computer simulation is like 0.01% of a potential solution.
Like I said, I’m not the person to ask and I’m still learning so I would be happy to be corrected on any of this.
odlird #31: believe what you want. if you think I’ve said anything unfair about these companies I’m happy to discuss.
I started interacting with QC investing twitter when my timeline started getting flooded with $ionq around August/September. Mostly because I’m in awe of how poorly these people understand what they are investing in. My colleagues I’ve shown posts to are also blown away by the willful ignorance of that community. I think more scientists should be aware of these things and how their work is indirectly being packaged and sold to investors.
Comment #33 January 12th, 2025 at 7:15 pm
I don’t get what all the fuss is about. My small, too recent, position in QTUM (a QC oriented ETF) was off 4.5% last week. Time to sell all? No way. Time to buy more? Perhaps. The log-log plots do look interesting.
Comment #34 January 14th, 2025 at 10:52 pm
I think it is impatient to understand why he said as so:
Nvidia is way behind on quantum chips, so it is natural for him to hype the stuff he has and down play what he does not have.
Comment #35 January 15th, 2025 at 12:10 am
Most AI companies shouldn’t worth anything either but he keeps making very bold claims about them.
investors are not looking at these as if all of them will succeed, however they are looking to bet early on one in a hundreds gems that is they succeed they would have one in a thousand profits.
Being able to just simulate quantum physics will open the door for a lot of technological absences in material sciences, drug discovery, …
Comment #36 January 15th, 2025 at 4:36 am
One reply to the NVIDIA comment is here: https://www.forbes.com/sites/jasonsnyder/2025/01/09/quantum-is-decades-away-not-so-fast/
I realise you’ve commented in depth on some of D-WAVE’s claims before. As far as you know, is there anything to the claim that quantum annealing is *already* being used to solve practical optimisation problems that wouldn’t otherwise be tractable (the part starting “My perspective is not theoretical.”) or is this just hype?
Comment #37 January 15th, 2025 at 7:27 am
Since this is the one chance I will get to be pedantic about pirate speak on this blog: avast means ‘stop’ or ‘halt’, not ‘vast’
Comment #38 January 15th, 2025 at 11:00 am
flergalwit #36: I think the evidence is flimsy that you’re getting any advantage over classical, with any problem of practical interest that can currently be solved using quantum annealing.
Comment #39 January 15th, 2025 at 4:15 pm
Jelmer Renema #37: Arr, ‘twas a pun, me mate! 🙂
Comment #40 January 15th, 2025 at 4:33 pm
“After all, even if there’s only a 10% chance that something will produce $100B in value, that would still justify a $10B valuation.”
This is actually being pretty generous. If you consider things like the time value of money and the risk premium, a 10% chance of getting $100 billion in 20 years is probably worth like $2 billion today. Though I think your basic point that it’s hard to assign a meaningful value to QC companies still holds.
Comment #41 January 15th, 2025 at 6:34 pm
In a similar vein re flergalwit #36 and Scott #38 and but this time from the CEO of a QC start-up (July 7 2024).
-Asked how SQC was going in the race to be the first to build a fully useful, error-correcting quantum computer, Simmons said the focus at this stage was on such “intermediate products” as the quantum AI accelerator, which is a piece of hardware designed to enhance the efficiency of AI applications and algorithms, for faster processing, improved accuracy, and reduced energy consumption. You use the quantum states to do that, so it’s directly related to the quantumness of the processor,”
What does “quantumness” mean here? (manipulatable interference? nanoscales? does it matter?)
and then, six months later (1 Jan 2025):
Q: And how close are we to the holy grail of the quantum computer?
A: Honestly for me our company is very excited. We predicted our first kind of products would be out in 2028 we actually brought them out last year, we’ve been trying them with customers already looking at applications such as fraud detection with the Commonwealth Bank, looking at how to optimize signals with the Telecommunications Institute with Telstra, looking at transport for NSW how to optimize traffic flow across the Sydney Harbour Bridge so we are now seeing applications becoming real-time for us so it four years ahead of schedule so it’s a super-exciting time for Australia.
You’ve previously emphasised:
Scott #14:
If someone invested in quantum computing specifically because they believe it’s poised to revolutionize optimization, AI, finance, and other broad classical domains in the near future—in that case, I strongly advise them to sell. We actually know something about these matters, and what we know directly contradicts what many QC startups have been telling investors and the public.
Is SQC an example of this dictum, a counter-example or neither?
Comment #42 January 16th, 2025 at 2:28 am
Scott #28
Dear Professor,
LIGO however did not have to be commercially useful. Even if we get a full-fledge QC, we have to find a relevant business case to justify the cost of running it (and developing it). One similar example might be nuclear fusion. We all know how it’s going: “it’s 20 years away and will always be”. But if it were to work, we know the payoff would be huge.
I love everything Quantum but I am worried, like many, of what this is doing to the science reputation, as if we weren’t in a lot of trouble already.
In any case, thank you for the time you take in writing this blog.
Matteo
Comment #43 January 16th, 2025 at 4:29 pm
Off-topic, but I couldn’t resist asking your take on this honestly disgusting article that came out in the Guardian: https://www.theguardian.com/commentisfree/2025/jan/16/i-knew-one-day-id-have-to-watch-powerful-men-burn-the-world-down-i-just-didnt-expect-them-to-be-such-losers.
Is it just me, or does this read like classic high-school-level nerd-shaming? Maybe I’m just too sensitive to such things.
Comment #44 January 17th, 2025 at 4:37 am
Scott #20: > “while I discounted the hype-addled crazies who claimed it might be a mere decade or so (!!) before AI would start reshaping the world in the ways foreseen by Turing and Asimov and Arthur C. Clarke”
Well, hoping to get R. Daneel Olivaw or HAL out of LLMs still seems to be kind of… crazy?
And while there are many claims that LLMs are passing the Turing test, it looks like nobody is actually doing “the” original Turing test aka Turing’s imitation game (if I am wrong about this, I’d love to read about one such test).
For reminders: Turing’s imitation game consists of 3 players. Player A is an AI tasked to convince the interviewer that it’s human. Player B is a human tasked to convince the interviewer that they are, well, human. Player C is the human interviewer and interacts with A and B simultaneously but separately through text messages. At the end C has to decide who’s A and who’s B.
Now we have to reward player B and C – they only get payed if they succeed. And maybe we can tell player A that we’ll shut it down if it fails. So everybody is incentivized to do their best.
Now we can repeat this a thousand times and look at a very simple statistic, just two numbers which will always add up to thousand.
That would be way more interesting than anything that people nowadays call “a” Turing test, which is basically play some input/output games with the chat bots and then build an argument around the output being indistinguishable from a human player.
Comment #45 January 19th, 2025 at 10:56 am
Kyle #43: People should feel free to criticize Zuckerberg and Musk and Bezos all they want for whatever bad things they do with the considerable real power they have, or (eg) for their craven appeasement of Trump. But when it crosses the line into attacking them for their real or perceived male nerdiness or social deficits—the trouble is that then you’re effectively attacking everyone else with those same traits or deficits. Now you’re punching down even while you pat yourself on the back for punching up. Reading the piece you linked, yes, it does appear to cross that line in places.
Comment #46 January 19th, 2025 at 8:15 pm
I’m sorry, what did Zuckerberg (and Bezos) do that’s a “craven appeasement to Trump?”
Comment #47 January 19th, 2025 at 8:40 pm
Democracy is a social invention that is as good as it gets for organizing human societies, but is no match for the evolutionary forces that appeal to our tribal roots. And as people like Yuval and Ian Bremmer have warned, the overlords of the algorithms(aka the centi billionaires) that exploit our evolutionary limitations will be the next superpower. Just look at the drama surrounding Tiktok, why would the previous regime intentionally make the law take hold just a day before the next regime takes over if there was really an intention to ban it ? It seems people on both sides of the aisle were just misleading the populace, national security be damned, or was it some sophisticated game theory maneuver ?
Comment #48 January 21st, 2025 at 10:54 am
Hey Scott, just getting around to watching your Bankless episode now. I took exception to this section, where you said:
> … if you got to a world where just about everyone had access to a quantum computer then it’s kind of amusing what would happen, which is that the proof of work just has its hardness just set automatically based on how much mining people have been able to do recently, and so all that would happen would be that pre-images would have to satisfy an ever more stringent you know uh condition and the proof of work would automatically just be made harder to compensate for Grover’s algorithm and we would all just be back where we started
Unfortunately, I think the situation is much more complicated than this – many people get this wrong. A key difference with Grover mining relative to classical mining is that a Grover miner does not complete a block until they measure their Grover register. This leads to a situation where the incentives are crucially different: When a Grover miner sees a new block, their incentive is not to immediately start mining on top of it, but rather to measure their own register and potentially publish the resulting block if it is valid. This can increase the so-called “stale rate”, which in turn undermines the security of the chain. I encourage you to read this paper [1] which analyzes this possibility, really there is a whole fascinating literature on Grover mining if you want to look into it.
[1]: https://arxiv.org/abs/1804.08118
Comment #49 January 21st, 2025 at 2:45 pm
@Dr. Aaronson #45
I will not dispute your opinion that the article may have crossed a line. But, I also remember the 1980’s when professionally employed women at a variety of incomes spoke of “not settling.” I am not convinced that every woman with such a view had been “nerd bashing.”
The author of that piece had been specifically referencing cringe comedy. I think that needs to be taken into account.
For most of my adult life I have worked in construction with men who have minimal education. My academic career ended because of illness (which is not to say it would not have ended because of poor academics at higher levels). Except for a handful of my coworkers, my interest in mathematics, computation, logic, and foundations had been identified with Sheldon from the show “Big Bang Theory.” By contrast, the others viewed that same interest analogous to Charlie from the show “Numbers.”
Clearly, television shows like “Numbers” or “NCIS” portrayed nerdiness of its more intellectual characters without the antics of cringe comedy.
Since I don’t know Musk or Zuckerberg, personally, I cannot say if their social media presence is performative or authentic. But, I can understand how a woman who thinks that societal leaders ought to behave with some dignity would be critical of men acting like they belong on a cringe comedy show.
The “father of modern economics,” Adam Smith, opined that a people will emulate the behavior of their leaders. Whereas I will acknowledge some feminist bias in the article, I do not necessarily think that concern for transforming governance into a comedy show should be seen as nerd bashing.
Comment #50 January 21st, 2025 at 10:35 pm
Ronald #10,
“Is Trump considered so much better for the prospects of quantum computing than the Democrats?” The $500 Billion in AI investment dollars have to go somewhere… Not saying it’s the ‘right’ connection scientifically unless we’re talking about the science of how money moves.
Comment #51 April 13th, 2026 at 5:28 am
I winced at the idea that Jensen Huang’s offhand “20 years” could knock 40% off IonQ/Rigetti/D-Wave in a day. The Satoshi hoard angle was new to me—the early, not-hash-protected coins—and it makes the “gold doubloons” analogy feel uncomfortably concrete.
Comment #52 April 27th, 2026 at 5:13 am
I’m struck by how Jensen Huang’s 20-year practicality estimate crashed markets for IonQ and Rigetti on just one analyst call. The line about Satoshi’s untouched bitcoin stash being potentially stealable by quantum tech makes me rethink crypto security more urgently than any academic paper I’ve read.
Comment #53 May 7th, 2026 at 4:59 am
It’s an interesting point about Satoshi’s untouched Bitcoin, and it highlights how quantum computing’s threat might materialize sooner for specific cryptographic vulnerabilities rather than a general market crash.
Comment #54 May 22nd, 2026 at 6:03 am
I’m struck by Jensen Huang saying quantum computing is twenty years away, and yet the market drop after that claim shows how sensitive asset prices are to one CEO’s estimate. The bit about Satoshi’s stash and elliptic-curve crypto adds a wild stake to the timeline debate.
Comment #55 May 25th, 2026 at 5:44 am
I was struck by Jensen Huang saying quantum computing is still twenty years away, and then seeing so many QC stocks drop 40% on that. It’s wild how a single offhand chart moment can swing billions, while I’m still curious about what “a scalable quantum computer” actually needs.