Multiple companies focus on qubits as next computing wave, but problems remain.
The race is on to commercialize quantum computing for everything from autonomous vehicles to supercomputers for hire. IBM has been working on a 50-qubit computer. Intel and QuTech, its Dutch research partner, showed off a 17-qubit test chip last month. And Alphabet, Google’s parent company, is developing a 20-qubit computer.
These numbers sound paltry compared to the billions of transistors being packed into existing processors, but even a 10-qubit computer leaves billions of transistors in the dust in terms of raw compute speed. This is what has the cyberwarfare experts so worried. The adage that, with enough time, any security can be cracked no longer holds with quantum computing. Given enough qubits, any security can be cracked quickly.
Still, the real money is in the commercial sector, providing a couple of big problems to be solved. First, most of the success has been achieved in the single digits of the Kelvin temperature scale. Researchers in Switzerland, Australia and Germany addressed this issue a couple years ago, showing it’s possible to extend electron spin lifetimes to 170 nanoseconds by using a film of burnt naphthalene particles (the stuff in mothballs) dispersed mixture of alcohol and water. That’s a long enough time period to do some calculations, but for commercialization it will have to last for microseconds to seconds.
Volkswagen and Google announced this week that they are working together on quantum computing. The goal is “practically oriented research” involving traffic optimization, new materials structures and AI with new machine-learning processes. Whether that occurs at extremely cold temperatures isn’t clear, but making cryogenic cooling portable would require more power than could be supplied by today’s batteries. It’s much easier to cool quantum computers in a data center than in a car, particularly when the computers themselves are so much more powerful than today’s servers, and that could have a big impact on what data gets processed where in autonomous vehicles.
There are other benefits to cryogenic computing, as well. At extremely cold temperatures, common DRAM exhibits superconductor properties. So access to data can be blazing fast using existing or new technology.
The second problem is that quantum computing isn’t entirely accurate. So while computations are extremely fast, they need to be handled today like a distribution instead of a single number. Work is underway to improve the algorithms that run on quantum devices, as well as the chips themselves. On the chip side, the challenge is being able to insulate the spin of electrons, which are critical to the states of qubits as 1s or 0s.
Morgan Stanley issued an excellent report on quantum computing in August, detailing the issues and benefits in commercializing this technology. Among the markets that it could address are machine learning, searching big data and medicine and materials, which the report identified as hard problems for classical computing.
Fig. 1: Projected uses of quantum computing. Source: Morgan Stanley
Quantum computing has opened up a whole other dimension for computer science, setting off a global race to utilize this technology for a variety of real-world applications. What this means for the chip industry isn’t entirely clear. At the very least, it will provide a brand new opportunity for everything from tools to equipment as quantum chips are refined and applied to real-world problems.
As with existing chips, size, portability, power and cost will need to be brought under control. And while scaling may not reach the single-digit nanometer range for decades, engineering will happen at the quantum level with new atomic structures and materials. Still, the volume of material on this subject is growing exponentially and money is pouring into this field, signaling this market has moved beyond pure research. It is now a race to commercialization, for better or worse.
Related Stories
Toward Defining Qubits
IEEE working on standard lexicon to talk about next-gen computing technology.
System Bits: May 30
Diamond-based quantum computing
System Bits: May 16
Quantum-circuit refrigerator
Manufacturing Bits: May 9, 2017
China’s quantum computer
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Ed, thank you for the article, yes, the field of Quantum computing is moving fast despite the enormous challenges which lie ahead in terms of commercialization. The investment is huge, and will grow, despite the reality that many approaches to harnessing quantum physics for the purpose of computation will fail. What many of these companies don’t realize is there is a new emerging technology called “modular computation”, which can be implemented in CMOS technology, and can accelerate AI and CNN’s by 10 to 50 times. This approach is new, and involves operating in a new number system, and so has been ignored by semiconductor companies at this point. Funny thing is the investment for modular computation is much less, and will provide guaranteed results in a much shorter time frame.
Looking at the Morgan Stanley chart, I would bet against it a bit. I’m an engineer whose seen the progress of technology, and while we talk about how fast things are going, we often don’t talk about how much time it takes to commercialize. 2025 will not bring Quantum computers that can displace anything standard computers can do. I don’t see a negative impact in high end computing supply chains as a result of quantum computing. Supply chains for high end computers will be affected by the slowing of Moore’s law … that’s the main culprit. Quantum computers are good at problems that conventional computers cannot do. However, what is NOT talked about is that Quantum computes cannot do much of anything that a standard binary computer can do! Increasing Qubits alone does nothing. Qubits must be set into a quantum system, a system that represents the problem to be solved. The Qubits must then co-mingle, and if noise and quantum decoherence doesn’t swamp out the computation, out pops an answer … in most cases we don’t even know if the answer is correct! Often times we need a binary computer to back track the solution, and determine if its even correct. Quantum computers guarantee no result, and may even produce many wrong results before a correct one is made. The reason is that Quantum computers are not deterministic, they are quantum. So yes, Quantum computing supply chains and technology will increase, but it will have little to do with affecting the digital computer landscape by 2025. But huge money will be spent, that’s for sure. Eric.
One last thing. That’s not to say I’m not a champion of Quantum computing efforts! I place in high regard researchers and companies willing to be the pioneers of a new method of computing, and I will continually applaud their efforts and progress!