Unpacking quantum myths...and why they matter

My inbound deluge of quantum computing press releases suggests that quantum computing might power enterprise apps within the next year or two. The hype suggests quantum will do everything better and faster thanks to the rapidly growing qubit count and new quantum algorithms. Well, not quite, suggest experts at Capgemini, during a conversation with me at Cambridge Consultants (part of Capgemini) headquarters.

But don’t be discouraged about looking more deeply just yet.  Quantum must be considered a complement to the existing IT infrastructure rather than a replacement. Much work must be done to set up the data, IT infrastructure, and new processes that can take advantage of quantum.

Julian van Velzen, CTIO & Head of Capgemini's Quantum Lab, said 2025 is increasingly becoming the year of quantum, even though practical hardware is still a few years out. The big immediate deal will be upgrading all of the cryptographic software protecting long-lived secrets and hardware that is more cumbersome to replace.

There is also an opportunity to reengineer business and research processes to be ready when quantum does arrive. For example, Airbus is already working with Capgemini to rethink its rust maintenance program to take advantage of better quantum simulation. van Velzen says:

We appreciate that these issues are not solved overnight, so it needs a longer perspective. But also, there is massive potential that depends on the compute power that you have available and the promise of quantum computing to really contribute to that. We appreciate that this will not be addressed overnight, but if you have the stamina to have a longer view, you can really change how you do these things.

Here are several common myths the Capgemini team sees.

Myth 1: quantum computing is too far out for planning

Quantum computing has been just around the corner for so long that it can be difficult to take it seriously. Now, IBM’s first large-scale fault-tolerant quantum computers are expected to begin running by 2028. van Velzen said that for companies like Airbus, which are working on a timeframe of thirty years to design the next plane, a few years might not be that long. Similarly, those timelines are within one or two development lifecycles in pharmaceuticals.

James Cruise, Head of Quantum Algorithms at Cambridge Consultants, says:

Those organizations that are really forward-looking and thinking about the next step for them are starting to engage here. And for example, organizations like Airbus are really starting to invest in this, and saying, ‘actually, what will it mean for us, and how will it change what we do,’ but also ‘what is the company change we need to put in place to enable us to do that.’ There are also huge challenges related to the integration with classical systems and classical processing.

Myth 2: quantum computing alone will be useful

Quantum computers can do some amazing things that will always be beyond the reach of classic computers, but they do other things slowly or not at all. Think of it as the difference between a slow-moving cargo boat and a freight truck. The truck will be faster. But the boat can cross channels. Cruise says it is important to think about how these can work together as part of an end-to-end system:

Quantum computing alone is not going to be useful. You need to use quantum computing in a holistic system with classical computing, because it's about finding the small calculation, which we can't do, and we can't imagine doing at the moment with classical computing that a quantum computer unlocks. How does unlocking that one calculation feed through the wider workflow and the wider chain of value? Unlocking that takes time and effort, and if we don't start working on that now, we won't be able to do it by the time we have these machines in 2030.

Myth 3: it's time to experiment with quantum POC

Cruise said it’s wise to avoid building quantum computing proof of concepts at the moment. At best, you might be able to test out the feasibility of a toy algorithm, but not how it works at scale. He suggests a more practical strategy is to focus on building the supporting classical infrastructure and processes. For example, in drug discovery, teams can start to use classical methods to look at slightly smaller examples using classical machines to plan the supporting infrastructure.

Another example is that Airbus has a finite element model that runs on classical methods. This can help them baseline the target accuracy they will need to deliver for a quantum computer to be useful.

Myth 4: quantum computers are faster than classical computers

Quantum computers will perform some types of calculations that classical computers will always struggle with. However, the physical speed of quantum processing is still much slower than can be achieved with classical computers. IBM’s roadmap plans to increase quantum operations from five thousand this year to one billion in 2033, but glosses over how much time this would take.

Cruise elaborates:

The bit which isn't talked about so much by the quantum industry is the speed of doing these operations. Because, as you rightly point out, there are sort of two parts to the statements. One is how many operations can I do before my calculation has too many errors to be useful, and that's often the number that IBM will quote, and that's the number that is talked about.

But there is a second issue, which is how fast you can do these operations, and these are then governed by the physical properties of the system that we are talking about, as well as different modalities. Different types of quantum computers have different speeds. The photonics are the fastest, but they're not so well realized at the moment. Superconducting ones are the most realized system, and they are looking at on the order of the megahertz scale, somewhere between ten and a hundred times slower than a GPU.

It's important to note that these numbers represent the physical qubits. Organizing these into logical qubits required for useful computation adds additional overhead, reducing performance by a hundred to a thousand times.

Myth 5: all quantum computing challenges are new

Researchers and vendors are increasingly finding that classical computers can play a supporting role in mitigating some of the limitations in quantum computers. For example, IBM is exploring how dedicated classical chips and GPUs can help support error correction processes. There are also opportunities to learn lessons from the early days of classical computer architectures.

Cruise explains:

Some of the challenges we see in quantum computing are actually challenges in classical computing ten to thirty years ago, particularly with things like non-interruptible processing. The depth of expertise that Cambridge Consultants has enabled me to talk to people who thought about those challenges then, and are able to help us build on those challenges now. Similarly, bringing in the expertise from AI, for example, can look at challenges of using AI to enable and build on quantum computing, bringing in that depth of skill base to really allow us to push the system forward and build a capability, not for today, but for that five to ten year time horizon.

My take

Enterprises should all start upgrading their cryptographic infrastructure today to prepare for quantum computing risks in the near future. I have had my doubts that it made much sense beyond that for most enterprises.

Cruise and van Velzen make some good arguments that getting ready for quantum is not just about installing the machines when they are realized, but planning the supporting processes and IT infrastructure. Even though the first quantum computers might be a few years out, there could be long-term value in laying the foundations for this today.