How much would you pay for a quantum computer?
Of course, it depends on exactly what this quantum computer can do, doesn’t it! If I give you a two qubit quantum computer, you may not want to pay me more than two bits (25 cents people, not two binary numbers.) Which is not to belittle the experiments that have been done to date which are few qubit quantum computers…these are among the most impressive works of experimental physics/engineering around. But I certainly wouldn’t pay much for the computational power these experiments demonstrate.
There are sort of two regimes where I think someone might actually want to buy a quantum computer. The first is when a quantum computer with around 100 qubits or so which can process some thousands of parallel operations before the computer decoheres/errors. Why would I be interested in such a machine? Well because I have no idea how to efficiently simulate some quantum systems of this size. Why do I go up to 100 qubits and not as some smaller number like 20 or 30. Certainly simulating quantum systems of this size is difficult. However, the systems which we would really like to use a quantum computer to simulate, those with a large amount of entanglement, are probably two (or higher) dimensional systems, and getting to a two dimensional system of ten by ten seems like a regime where I can at least begin to rid myself of some small finite size effects.
The next step, of course, is a full scale quantum computer, one which is operating below the threshold for fault-tolerant quantum computation. What price should we assign such a device. Again it depends on the exact specs. But let’s just assume that this quantum computer has a few kilobytes of quantum memory. What will the clock speed of our quantum computer be? Well it will certainly depend on the physical implementation. And there is the overhead of quantum error correction. So the clock speed may range anywhere from MHz, to even PHz. How much would you pay for such a quantum computer?
For comparison, IBM’s Blue Gene, the worlds fastest supercomputer (that we know about) today, cost around one hundred million dollars.
Let the bidding begin!
The qBabbage: 100 qubit quantum computer, with the ability to perform, say 1000 operations before decoherence/noise ruins a quantum simulation. Start bids at 10 thousand dollars.
The qMark I: A fault-tolerant quantum computer with 2 kilobytes of quantum memory and a clock speed of MHz. Start bids at half a million dollars.
The qWhirlwind: A fault-tolerant quantum computer with 2 kilobytes of quantum memory and a clock speed of THz. Start bids at one million dollars.
O brave new quantum world!
What? No one wants to buy a quantum computer from me?
Well maybe if you were also offering access to say one of the transatlantic cables that carry IP traffic. Then I’d stand a good chance of being able to make the money back by playing with bank transfers.
On the other hand it seems likely to me that once we have quantum computers of a certain level of sophistication, then the speed at which the whole field progresses will increase dramatically. Once one can efficiently do quantum simulations, the need to wait for experiments to be performed reduces. So if we can try many designs for quantum memory or registers in a fraction of the time it takes for them to actually be built, then it seems likely that we will have to wait less time for a major success. We could also see the use of genetic algorithms, etc, improving on our basic designs.
With this reasoning in mind, I bid $10,000 ffor the qBabbage, and hopefully I’ll be the one selling the qMark 1 and the qWhirlwind.