Commercially Useful Quantum Dragons
But what are they commercially useful for?
The fine folks at Inside Quantum Technology were having a private conversation the other day, and I don’t think they’ll mind my letting you in on it. You see, we were going back and forth over the timeline for commercially useful quantum computing. Since NVIDIA’s CEO said it’s about 20 years away, we’ve had the industry celebrating almost every prediction that has been earlier than that. One of the most recent predictions has been 5 years, which coincides with the growing number of roadmaps that promise we’ll have logical qubits by 2030.
But then I remembered working on this:
If you’ve read the Quantum Random Number Generator Markets 2024 report, then you know that quantum computers are commercially useful today. They are commercially useful for a specific purpose, but they are commercially useful nonetheless.
Useful for what?
Instead of asking when quantum computers might be commercially useful, we should be asking when they'll be commercially useful for _____. We need to ask a full question, to which the answers ought to be a series of timeframes, not just one.
For example, they are being commercialized today for random number generation for a range of applications and sectors. To the opposite extreme, I’ve been told by multiple researchers that we’ll need way more than 100 logical qubits, extending well past 2030, before quantum computers will be commercially useful for chemistry and material science applications.
In between today and more than five years from now, what do we have? Optimization problems? Machine learning problems? We tend to group use cases into these broad classifications, but not all use cases are equally hard. The advent of utility ought to be staggered accordingly.
What will run where and when?
We not only have to ask better questions, but we also need to request substantive answers. In the year 2030, which algorithms will run on whose hardware and be commercially useful for solving which use cases? In the year 2035, same question. In the year 2045, you see my point.
We also need to define the term “useful.” Is it enough that we can solve a use case? If it’s not advantageous to do so, should it at least be competitive? Quantum computers are legitimately competing with other classes of QRNGs today, so why should the standard for computation be any different?
Conclusion
So, the next time someone says, “quantum computers will be commercially useful by _____,” your reply should be “quantum computers are commercially useful now, please be more specific.” If you feel the need to strengthen your argument, allow me to once again shamelessly plug the Quantum Random Number Generator Markets 2024 report:
Image generated by an AI model provided by Microsoft Copilot.