15. The Quantum Century.
The 21st century will be largely dominated by what happens in Quantum Computing - a full scale "fault-tolerant" quantum computer may just be Mankind's most profound invention yet.
Quantum Computers are fundamentally different. The word “computer” in the name is probably to blame. The uninitiated would think that QCs are merely the next evolution of computing. They are not.
A more apt analogy1 would be to think of QCs as the warp-speed travel machines in sci-fi movies which allow for inter-galactic travel. Compare that to our bikes and cars which are akin to today’s computers (even super-computers). Yes they are alike in that they get you from A to B, but are they really alike?
To understand how QCs are different, let’s first briefly look at how a normal “classical” computer works. They work by using bits to perform calculations, that are represented logically by either a 0 (off) or 1 (on). Computers have evolved to be very powerful over the decades on the back of Moore’s Law, but the fundamental science remains the same: all computation is done by bits that are performing lots of these 0s and 1s calculations.
QCs are fundamentally different because they use the superpowers of quantum-mechanics: Superposition and Entanglement (do read the previous post for background). Enter Qubit.
No, that’s the wrong Qubit. That is my Mother-in-Law’s puppy, also called Qubit (you can guess I spent a lot of time discussing quantum in her house during the lock-down!).
The qubit I am referring to here is the “quantum-bit” that is used in a QC. Thanks to quantum superposition, a qubit can exist in both states 0 and 1 simultaneously. This makes qubits far more powerful in their ability to store and calculate information than classical bits that need to process each calculation as a binary.
To make qubits work together, we need to entangle them. Once entangled, two (or more) qubits can no longer be described independently (remember they now have the same wave function). The result? Exponential increase in computational power as you increase the number of qubits. Even with 65-75 high quality (“logical” or even “algorithmic”) qubits, you could run algorithms in seconds that would take even the world’s best supercomputer billions of years to compute.
The magic of quantum mechanics allows for this exponential power. Two entangled qubits in superposition have four possible states (2^2 = 4). Three have eight (2^3 = 8). Each time you add a qubit, you double the power of the QC. Exponential. Exponential as a concept is almost incomprehensible to the human mind. 10 qubits have 1,024 possible states (2^10). However, you scale that up to ~300 qubits, and you have more possible states than there are atoms in the known universe.
Remember how Feynman said that nature is quantum (and exponentially increasing in complexity) - well we may soon have a machine that can match it. It’s hard to exaggerate how profound this is.
A small digression before I touch on some of the use cases of this computational power. If you took all the atoms in the universe and made a supercomputer out of it, and gave it the age of the universe to compute (~14 billion years), it still wouldn’t be able to do calculations that QCs could. How can this be possible? One theory suggests that QCs are drawing their power from parallel universes. The theory suggests that a qubit in superposition is actually 0 in one universe and 1 in another, and so on for more entangled qubits. This theory is by David Deutsch, a physicist at University of Oxford, and is a branch of the Many Worlds interpretation of quantum mechanics. I find it totally fascinating. Check out this Joe Rogan interview with Sean Carroll for more on Many Worlds.
Back to our universe (or is it). It is impossible to imagine today what all the use cases will be for powerful QCs. The problems they are most suited to solve have a common quality: combinatorial explosion, where there are many different variables with one optimal solution. The best systems today have 10-20 logical qubits. Really interesting things can start to get done at 50+ logical qubits. Of course, QCs of this power do not exist today, but they are coming fast (5-10 years depending on published roadmaps). Do not underestimate how quickly exponential increases become unstoppable. Some of the use cases one can think of:
Quantum Simulations in Chemistry: In the future, people will wonder how chemistry was done at all without a QC. Most chemistry labs will be replaced by QCs, and labs will only be used for the actual synthesis. Think of it as making airplanes today without computer models. Today’s planes go through 100,000s of versions on computer models before they are put in production. Yes, the Wright Brothers flew without computer models. That’s how today’s chemistry is being done without QCs. Rather than synthesizing every molecule and seeing the results, a QC will be able to tame the combinatorial explosion in molecular design and find optimal solutions, before they are sent to labs to synthesize. This will dramatically cut the time and resources for everything from drug discovery, to inventing new materials, to improving battery power.
Once you can simulate nature, we can learn from photosynthesis and improve photovoltaic solar cells.
We can learn more efficient fertilization from microbes/enzymes and finally replace the 100+ year old Haber-Bosch process for fertilization, which is responsible for ~2% of the world’s energy consumption (trivia: did you know that roughly half the Nitrogen in your body was made in a factory using this process… maybe less if you have a more organic diet)! We have no hope of simulating this with a classical computer (unless we get randomly lucky), the combinations are just too complex.
Logistics: Companies will be able to save a tremendous amount of money by optimizing the routes their drivers take. If the average driver makes 120 deliveries a day, the total possible combinations he/she takes is a number with 199 digits. It’s estimated the company could save $30 million by figuring out how a single driver can cover one less mile a day. Multiply that across its entire network, and the savings could be staggering2.
Climate Change: We could figure out more efficient ways to capture and remove carbon and other greenhouse gases from the atmosphere.
Quantum Algorithms for Monte Carlo Simulations: These are probability simulations to calculate the expected distribution of possible outcomes in processes involving random variables. Many industries use these, with finance being the most obvious. More broadly, once someone (say a prop trading desk at a bank) in finance starts using a powerful QC for applications such as options pricing etc., their competition don’t stand a chance. This can probably be extended to most industries at some point: once your competitor has found a way to smartly use a powerful QC, you have to do the same, or perish.
Machine Learning and Artificial Intelligence: A QC can “learn” things that are simply beyond the capabilities of classical computers. It is impossible to imagine the implications of combining a QC with AI/ML.
Cryptography: This is probably why governments (especially China and the US) are so keen to develop QC. A powerful QC will be able to crack the most sophisticated of today’s encryption (i.e. RSA encryption) in seconds/minutes. That could compromise the entire Internet. A nation’s entire military could be taken off-line in a few minutes. Happily, this is still several years away. But preparation should start now.
There will be many more applications that we can’t even imagine yet.
I would think you are now convinced of the world changing potential of QC. So what should we do about it? If I was a bit younger, I would try and join a QC company. What I am actually doing is a lot of research on the best QC companies out there; to invest in them given I can’t work in them (or start one!). I think there is a good chance there will be some multi-hundred billion (maybe trillion) dollar market cap companies in this space in the future. I will present some thoughts on this in the next post.
Analogy borrowed from Jeremy O’Brien, CEO of PsiQuantum