[00:00] Quantum computing has a deadline. Following  an executive order of President Trump,   the US-American Department of Energy  has announced that they will “deploy the   world's first fault-tolerant, scientifically  relevant quantum computing capability for   [00:16] research and development by 2028.” Awesome!  Can we please also have an executive order   for quantum gravity by 2030, and while  we’re at it, immortality by 2040? Yeah,   [00:29] I think it’s not going to happen, and today I  want to tell you why. Stay around until the end   for a special offer that will bring your coding  skills to an entirely new level with boot.dev. [00:41] For years I have told you that I think  quantum computers are not going to scale   and now the authors of a new paper finally put  the numbers behind it. It’s the first attempt   to calculate the physical resources that  a large error-corrected quantum computer   [00:55] would require. They look specifically  at superconducting circuits as qubits,   which is the most studied and  most advanced current approach. The problem with quantum computers  is apparent if you just look at   [01:08] this footage. This thing which you see  here has little to do with computing,   it is mostly a noise buffer and equipment  to cool a tiny microchip at the very bottom. These chips may contain 1,000 or maybe  a few thousand of these qubits at most   [01:24] before they simply get too large for the  cooling apparatus. If you want to build a   serious quantum computer that can do  all the things we’ve been promised,   you’d need a few million qubits. Then the  problem becomes you have to combine multiple   [01:38] of these cooling devices. Leaving aside  that no one really knows how to do this   without ruining the required entanglement, it  becomes a problem simply because of the size. [01:50] The authors of the new paper say, let’s  look at something like 4 million qubits,   which is a reasonable estimate, and be generous  and say that we can put about 10,000 qubits in one   cooling apparatus. The entire quantum computer  would then cover an area of more than 800 m²;   [02:08] that’s about the size of a German Aldi  supermarket. And that is without any of   the necessary equipment to manipulate and read  out the thing. Add all that equipment on top and   we’re talking about at least twice the size, not  counting all the extra facilities necessary to   [02:24] build the components. So if you were thinking we’d  have a factory that produces quantum computers   like laptops, it’s not going to happen. The  factory will be the quantum computer, basically. And with a size like this the noise  problem is going to explode. I am sorry,   [02:41] but really I see no way you will ever  error correct anything that size. Now look,   there are other approaches to quantum  computing that use other technologies,   like for example PsiQuantum that wants to use  photons as qubits. But no one has even seen a   [02:55] demo of the thing. In what I am sure are  completely unrelated news, top executives   of leading publicly traded quantum computing firms  have sold off significant numbers of their shares. It is not that I am somehow against quantum  computers, I am just extremely sceptical   [03:11] of their supposed practical uses besides the  research opportunity of just studying a big,   highly entangled state. And keeping these  researchers employed. That, to be fair,   is a quantum computing application  with excellent experimental support. [03:27] Let me give a few timely examples to explain why  I am sceptical. In 2023, the American non-profit   Wellcome Leap offered $5 millions to anyone  who would find a healthcare problem where a   [03:41] quantum computer could beat a conventional one.  They published the results in April. The winner   was… no one. No one could find a good use  case for a quantum computer in healthcare. [03:53] This isn’t an isolated case, I have  previously reported that whenever someone   looks closely at what a quantum computer  can supposedly do the answer is not much. In a recent paper, a team from Los Alamos looked  at the question of whether quantum computers are   [04:10] actually good to speed up neural networks,  “quantum AI” as the headlines like to put it.   They found that the conventional algorithm  would match or beat the quantum version on   every benchmark they tried, up to more  than1,000logical qubits. So if you want   [04:26] to make quantum artificial intelligence more  efficient, just remove the quantum computer. The authors say that it might be that there  are some sorts of datasets for which quantum   [04:38] computers might have an advantage, but  that hasn’t been demonstrated and then   there is again the question of who would  actually have a use for that. Likewise,   salesman problem and related logical and  logistic problems or finance applications. [04:56] Now look. I openly admit that I am  cherry-picking bad news to confirm   my biases. But that doesn’t mean this  bad news doesn’t exist. The one so far   solid use case of quantum computers remains  cracking old ciphers. That is not nothing,   [05:11] but I think it doesn’t remotely justify  the investment. It pains me to think what   good all this money could have done had it  been invested in something that benefits the   public. Like science communication that could have  prevented this utter insanity in the first place. [05:27] Artificial intelligence can do a lot, but  if anything, it's increased the demand for   software engineers who understand how to  code. And there's no better place to get   started than with today's sponsor, Boot.dev.  Think about everything you've done today,   [05:42] and chances are most of it was accessing some  sort of data through a user interface. Because   that's how most apps work, how most businesses  operate, and how most research is presented. And   [05:55] that's exactly what you'll learn on boot.dev,  backend development, database management,   user interfaces. They have hands-on lessons in  which you write code directly in the browser   [06:07] and you solve challenges to level up. And it's not  single scattered lessons. They have very carefully   designed learning paths that'll get you started  on anything from Python for beginners, Typescript   [06:20] to your first personal project. If you decide to  go for the full experience, use my code SABINE   to get 25% off your first year. So go and check  this out. Thanks for watching. See you tomorrow.