Soon to be followed a few years later with the same money pouring in to PR/consultancy report hype for quantum tech soon to be followed by a a predictable death for anyone dumb enough to have fallen for it whilst the startup investors divested long before.
I do not think historically this is what happened to the same type of hype for transitors/integrated-circuit startups a few decades ago (some of them failed, some of them reshaped both the technology and economy landscapes).
Certainly, the amount of hype and uninformed investing are damaging to the field, but coherent control of quantum states (whether for computing or not) is a "very big deal", that would create amazing technologies from sensing and metrology to (hopefully) computing.
This a tangent to the topic of the article, but it is an interesting tangent: There are many ways in which having coherent quantum states permits better sensors (sensing anything that is related to a Electric/Gravitational/Magnetic field, voltage, current, intensity, frequency, so basically all kinds of areas).
- coherent superposition of multiple sensors improve the signal to noise ratio
- improving beyond the diffraction limit of telescopes, by quantum-entangling multiple sensors
- using constructs similar to quantum error correcting codes to make measurements more resilient to noise
And these above are just the generic suggestions that are discussed in grad classes. When you dig into it there are so many other amazing quantum tricks to do.
My point is that this particular technology is very much worthwhile, even if some of the startups are making annoying and misleading claims (DWave for instance, which indeed sells expensive devices that are not quantum computers and do not seem to be that useful). As the technology is developed, the misleading claims will simply fall behind and be forgotten.
Well when that was said about the internet, they were kind of right. I don't know how you would believe quantum computing wouldn't be a huge change unless you thought something like error detection made it fundamentally impractical.
It's going to take decades to mature the technology but it will change a lot. As an investment tool, it's much less certain. Capitalism is fluid. If money is easily available, people will start bad companies to get that money. The internet went through a bubble but certainly did change everything.
No one has said anything like that. Not even the more "media-savy" startups. Everyone is talking about how on very specific problems you will get some sort of time or energy advantage.
It's been quite a long time since capital actually invested in things whose goal was to deliver something and make money by selling it. These days most capital chases hype with the goal of cashing out before the hype collapses.
This works because we have a severely demand constrained economy with monstrous amounts of money sloshing around at the top with no idea where to go. The idea is to get a hype train rolling to scoop up some of it. Nothing of value has to be produced.
It's just another field with some interesting (modest) applications but where 99% of startups are likely either knowingly peddling snake-oil or are total moonshots that have no chance of succeeding.
EDIT: Just to be clear: there are likely some interesting applications and some companies pursuing them might end up making it big, but the "gold rush" very much is mostly companies trying to ride on the hype until they crash and burn predictably.
When real world quantum applications are unlocked they have large commercial application. It’s too early to know if this wave of funding will get the returns it seeks but it’s still a promising sector.
I see universities trying to bolster interest in quantum computing by putting out open source projects to simulate programming for one. I suppose their hoping some kid will create their industry by giving them a cool idea.
I genuinely cant wait to see what will come of it, but I wouldn't be surprised if they're expecting quantum snapchat or quantum youtube.
thats what "gold rush" means. something useful and real, where MOST people who go after it dont get it, but a few do. the majority of people get caught in the hype and rush, and never see any profitable gold.
It's a bloody shame they're squishing in nonsense startups on quantum computing with useful imaging and cryptography startups. The latter are places where investment dollars are actually going to pay off.
Also kind of shocking how nature has devolved into a sort of science tabloid format. I suppose that happened a while back, but I remember when being published there in any format meant something.
Nature News is published as a column in the physical printed version of the Nature journal, no? So it's true it's not a Nature journal research article, but I think that's what scottloclin meant when he wrote "...published there in any format...".
There seems to be some disenchantment about quantum computing in the comments already, but I'll add my take:
This sounds similar to the AI craze that took off in the 80s (or was it the 70s?) where we were making such great progress that general AI was just around the corner. Tons of people got on the hype train and then we just never delivered. Investment dollars fell off, and we got the "AI winter." But then a few advances were made such as deep learning, and another craze has taken off with AI (now called machine learning) with slightly different expectations.
It sounds like quantum computing promises a lot, and we have mathematical proofs that show that a proper quantum computer could do great things. Right now we are making tons of progress, and money is flowing in. But eventually I think that progress will slow down considerably before we reach the level of every corporation having a general quantum computer in their server room, and investment dollars will fall off. Then a decade or two later another breakthrough will happen and the hype train will start rolling again.
Edit: I guess what I wanted to get at is that you probably shouldn't call people stupid for investing in promising technologies. The bubble will probably burst, but a venture capitalist probably shouldn't hold back their investment money because they think the "real" breakthrough will happen next decade. If real breakthrough is happening now, but the VC waits, they may have missed out on one of the most profitable opportunities of their lifetime.
Quantum computing may someday be amazing, but at the current rate of progress, in 5 years it'll still probably be a side show. There's a small-but-reasonable chance of some sort of massive breakthrough that will make them suddenly jump in power, but I wouldn't call it a good bet from a financial perspective. (Bear in mind, for instance, that while Google may have demonstrated a 53-qubit machine, it's still deeply limited and they had to construct a problem for it to demonstrate quantum supremacy on. It isn't even an arbitrary 53-qubit machine yet.)
At the moment, I see no clear path towards monetization for these startups yet, in any time scale that makes sense for a startup. AIUI it's not like there's even a ton of algorithms that you'd want to run on a quantum computer yet that are known to be clear quantum wins. For instance, breaking encryption requires a qubit per bit of encryption, IIRC; it's not just that "quantum computers break encryption" but "4096-arbitrary-qubit computers break 4096-bit encryption". 4096-qubit machines seem a long shot on the 5-year time frame, even interpreting current progress as exponential, let alone a 4096-qubit machine that can actually run Shor's algorithm.
Five years? I was involved in a graduate research program in quantum computing in 1996! It really hasn't gotten too far in all that time and I'm pretty happy that I didn't make that my dissertation topic because it was just not timely or possible to make decent progress on any of it back then. We're more than 20 years on and it's still dubious as hell.
Street cred: Steve Jurvetson has been investing in quantum startups since the early 2000s. It is discussed at the very beginning of his Tim Ferris interview: https://tim.blog/2018/05/31/steve-jurvetson/. Highly recommended.
HN commenters seems to have the amazing ability to predict the future with incredible precision. Why we need quantum computers if we have HN commenters?
The interesting and missing thing from a lot of discussions is what the investment bet is (beyond flipping before the musical chairs ends.)
AFAICT, assuming things keep progressing:
-- General purpose quantum computing is easily 8-10 years out: https://twitter.com/lmeyerov/status/1177632571636584448 . HW vendors might become lucrative around then, and soon after, SW. As a language & framework tool designer, I'll probably poke my head out again in a few years again.
-- Niche stuff, especially military, can be done over the next few years, with subsequent ROI in the 5-10 year range. Need funny secure radar tech with lossy analog circuits? Go for it!
-- Medium term: There may be some wins around say HFT doing probabilistic optimization algorithms on super lossy/unreliable quantum circuits, and maybe national label + pharma stuff for the same. This might be interesting for a larger-than-expected audience (breakthrough events for synthetic bio, genomics?), and as a tool designer, blaze the path to general SW, which is also interesting..
At this point in time there is no indication that quantum computers will ever solve a useful problem. Even if there is one it's not likely to be profitable.
25 comments
[ 0.19 ms ] story [ 65.1 ms ] threadCertainly, the amount of hype and uninformed investing are damaging to the field, but coherent control of quantum states (whether for computing or not) is a "very big deal", that would create amazing technologies from sensing and metrology to (hopefully) computing.
in what kind of areas?
- coherent superposition of multiple sensors improve the signal to noise ratio
- improving beyond the diffraction limit of telescopes, by quantum-entangling multiple sensors
- using constructs similar to quantum error correcting codes to make measurements more resilient to noise
And these above are just the generic suggestions that are discussed in grad classes. When you dig into it there are so many other amazing quantum tricks to do.
The same type of history had been identified as a way to get things sold.
You can buy the pr for any mediocre technology and it's success is irrelevant.
Nope.
It's going to take decades to mature the technology but it will change a lot. As an investment tool, it's much less certain. Capitalism is fluid. If money is easily available, people will start bad companies to get that money. The internet went through a bubble but certainly did change everything.
This works because we have a severely demand constrained economy with monstrous amounts of money sloshing around at the top with no idea where to go. The idea is to get a hype train rolling to scoop up some of it. Nothing of value has to be produced.
It's just another field with some interesting (modest) applications but where 99% of startups are likely either knowingly peddling snake-oil or are total moonshots that have no chance of succeeding.
EDIT: Just to be clear: there are likely some interesting applications and some companies pursuing them might end up making it big, but the "gold rush" very much is mostly companies trying to ride on the hype until they crash and burn predictably.
I genuinely cant wait to see what will come of it, but I wouldn't be surprised if they're expecting quantum snapchat or quantum youtube.
Also kind of shocking how nature has devolved into a sort of science tabloid format. I suppose that happened a while back, but I remember when being published there in any format meant something.
This sounds similar to the AI craze that took off in the 80s (or was it the 70s?) where we were making such great progress that general AI was just around the corner. Tons of people got on the hype train and then we just never delivered. Investment dollars fell off, and we got the "AI winter." But then a few advances were made such as deep learning, and another craze has taken off with AI (now called machine learning) with slightly different expectations.
It sounds like quantum computing promises a lot, and we have mathematical proofs that show that a proper quantum computer could do great things. Right now we are making tons of progress, and money is flowing in. But eventually I think that progress will slow down considerably before we reach the level of every corporation having a general quantum computer in their server room, and investment dollars will fall off. Then a decade or two later another breakthrough will happen and the hype train will start rolling again.
Edit: I guess what I wanted to get at is that you probably shouldn't call people stupid for investing in promising technologies. The bubble will probably burst, but a venture capitalist probably shouldn't hold back their investment money because they think the "real" breakthrough will happen next decade. If real breakthrough is happening now, but the VC waits, they may have missed out on one of the most profitable opportunities of their lifetime.
At the moment, I see no clear path towards monetization for these startups yet, in any time scale that makes sense for a startup. AIUI it's not like there's even a ton of algorithms that you'd want to run on a quantum computer yet that are known to be clear quantum wins. For instance, breaking encryption requires a qubit per bit of encryption, IIRC; it's not just that "quantum computers break encryption" but "4096-arbitrary-qubit computers break 4096-bit encryption". 4096-qubit machines seem a long shot on the 5-year time frame, even interpreting current progress as exponential, let alone a 4096-qubit machine that can actually run Shor's algorithm.
AFAICT, assuming things keep progressing:
-- General purpose quantum computing is easily 8-10 years out: https://twitter.com/lmeyerov/status/1177632571636584448 . HW vendors might become lucrative around then, and soon after, SW. As a language & framework tool designer, I'll probably poke my head out again in a few years again.
-- Niche stuff, especially military, can be done over the next few years, with subsequent ROI in the 5-10 year range. Need funny secure radar tech with lossy analog circuits? Go for it!
-- Medium term: There may be some wins around say HFT doing probabilistic optimization algorithms on super lossy/unreliable quantum circuits, and maybe national label + pharma stuff for the same. This might be interesting for a larger-than-expected audience (breakthrough events for synthetic bio, genomics?), and as a tool designer, blaze the path to general SW, which is also interesting..