Estimates of the commercial availability of quantum computing range from "right now" to "ten years." No matter when, one thing is certain: encryption is in peril. A quantum computer has enormous power to break anything digital.
All of the encryption models used today will be child's play for a quantum computer. That should keep all of us awake at night.
There are presently three types of quantum computers. We don't know what other types lurk in labs in China, the US, and other nation-states, but the following three types are well-documented:
Annealing Quantum Computer is the entry-level quantum computer. It is the least complex, and therefore the most straightforward to build. But it is also the least powerful. Many HPCs (High-Performance Computer, a.k.a. Supercomputers) working today can outperform an annealing quantum computer on typical tasks like video creation, gaming, email, etc. However, the Annealer can crush the performance of HPCs, factoring in very large numbers (a necessary feature in cracking encryption). An intractable problem for classical computers is solving optimization problems. Classic algorithms for route or mapping problems, or any optimization problem with many legs, are subject to the combinatoric expansion, which chokes classic computer architectures.
The jury is out whether Annealers will last. For a more deep technical description, see this PDF.
Analog Quantum Computer - seems to have the momentum in research dollars, and its performance is staggering. Google, Microsoft, IBM Q, Rigetti, Honeywell and IonQ have chosen to pursue this option, despite requiring far more precise and rigorous manufacturing. In theory, it can solve massive problems in quantum dynamics, material science, quantum chemistry, optimization problems, and sampling. It's demonstrated the ability to break encryption using algorithms, like SHOR's or Grover's.
Universal Quantum Computer - poses serious hurdles to build because it is so complex. It is leagues ahead of the others in terms of power. While current quantum computers are chugging along at 100-200 qubits, this monster will have more than 100,000 qubits, perhaps a million or more with qubit technology not yet invented. In an era when sustainability is stressed (the fastest HPC ten years ago consumed 40MW of electricity, the fastest ones now, even running 60-100 times faster, are running on less than 10 MW due to miniaturization). The more difficult problem is that it requires cryogenic cooling to absolute zero to operate. See my article: Quantum computing is right around the corner, but cooling is a problem. What are the options?
On orchestration (Hybrid Quantum)
One thing that puzzles me is there is very little available information about the orchestration of quantum computers. There are more things a quantum computer can't do than it can. The only company I've come across that discusses the quantum "stack" and what it looks like is Quantum Machines. This interesting Israeli company developed (and sells) a universal orchestration package that manages the quantum operations. They've developed quantum algorithms and even quantum programming languages. As Dr. Itamar Sivan, Co-Founder and CEO, Quantum Machines, explained to me:
Although reports in the popular press tend to focus on the development of qubits and the number of qubits in the current prototypical quantum computing chip, any quantum computer requires an integrated hardware approach using significant conventional hardware to enable qubits to be controlled, programmed, and read out.
Sivan feels that focusing solely on the number of qubits is just part of the equation. According to Dr. Sivan:
While today's most advanced quantum computers only have a relatively small number of available qubits (53 for IBM's latest generation and 54 for Google's Sycamore processor), we cannot maximize the potential of even this relatively small count. We are leaving a lot on the table with regards to what we can already accomplish with the computing power we already have. While we should continue to scale up the number of qubits, we also need to focus on maximizing what we already have.
Artificial Atoms - mad science, or not?
Put this under the category, "Is this weird or is it terrifying?" Didn't we learn it's not nice to fool Mother Nature?
In a paper published in Nature Communications, UNSW quantum computing researchers describe how they created artificial atoms in a silicon 'quantum dot', a tiny space in a quantum circuit where electrons are used as qubits (or quantum bits), the basic units of quantum information:
Scientia Professor Andrew Dzurak explains that, unlike a real atom, an artificial atom has no nucleus. However, it still has shells of electrons whizzing around the center of the device rather than around the atom's nucleus.
'But what excites us about our latest research is that artificial atoms with a higher number of electrons turn out to be much more robust qubits than previously thought possible, meaning they can be reliably used for calculations in quantum computers. This is significant because qubits based on just one electron can be very unreliable.'
So, the development of quantum computing continues apace, but what if artificial atoms obliterate humanity? Mad science.
There is a groundswell of opinion that commercially viable quantum computing is closer than anyone thought. I think this is an optimistic view. A reliable device based on quantum mechanics seems aspirational. Solving the problems of noise, fragility, and cooling to create a reliable quantum state, and then just the manufacturing technology needed, is daunting.
New materials are needed that haven't been invented yet to put one million qubits on wafer to solve the truly important problems. Nevertheless, computers with <100 qubits have been shown to have some potential. There are operational quantum computers already for public use. Amazon offers three quantum computers in the cloud.
- D-Wave - Advantage system is an annealing quantum computer with 24 qubits.
- IonQ is a universal gate quantum computer with trapped ions. More specifically, there are 11 qubits made out of Ytterbium with a trap-chain defined by a laser.
- The Rigetti is another computer with superconducting qubits. It has 30 qubits (more than IonQ).
Microsoft Azure, IBM and Google also offer quantum computing in the cloud. However, if someone predicted ten years ago that supercomputers would reach three exaFLOPS by 2021, I would have been skeptical.