By Yuval Boger

Full-stack vendors are quantum computing companies that provide both computing hardware and software. The potential advantages of an integrated offering were used, in part, to justify several mergers and acquisitions, such as Quantinuum (combining Honeywell’s hardware with CQC’s software) and Pasqal (a hardware vendor that merged with Qu&Co, a software company). 

Proponents of the full-stack approach believe that tight integration between the hardware and software teams can lead to better quantum software. For instance, an internal software team might be able to fully leverage a unique hardware capability such as Quantinuum’s in-circuit measurement or ColdQuanta’s global phase gates. The hardware team might be more open to discussing known problems and limitations of the hardware and work together to devise ways to bypass them. Furthermore, a captive software team might assist in specifying the requirements of the next-generation hardware platform since the software team knows what capabilities they need.

Apple is inevitably mentioned in these discussions. Not because Apple makes quantum computers – they don’t – but because the tight software/hardware integration has yielded a superior user experience for users of the iPhone, Mac, and other Apple products. 

Some analysts, on the other hand, might consider a full-stack company to be one with a lack of focus. It might be hard to fully fund both a hardware and a software program. Software innovations could come from dozens of vendors outside the full-stack company, and tying the software to the hardware might discourage these outside vendors from collaborating with the company.

And when the subject of Apple comes up, these same analysts would be glad to point out that it took Apple 30 years to reach this point and that, unlike some of the current quantum technologies, there was never a concern that silicon-based CPUs just won’t work.

Shareholders and analysts aside, what should customers prefer?

For customers that need an immediate solution and are willing to live with the severe limitations of today’s computers, a full-stack solution may be attractive. It’s a single point of contact, with little or no need for integration testing. But in my opinion, there are stronger arguments for “best of breed” components that do not necessarily come from the same vendor.

Why is that the case? It is entirely unclear today which technology, not to mention which vendor, will win the quantum race. Will it be superconducting qubits? Trapped ions? Optical qubits? The “best” quantum computer changes every other month, and a model that is best for one application might not be best for another. In my experience, customers are interested in “hardware portability,” in the ability to quickly move software from one platform to another. Customers prefer to benchmark their solutions on different hardware platforms, looking at the ability to fit the problem onto each platform, the accuracy of the result, and other parameters.

Furthermore, one could make the case that – even today – there are too many hardware companies and that some of them will not be around in a few years, whether because they could not secure enough funding, because they were bought out, or because they bet on the wrong technology.

Choosing a full-stack solution from a full-stack vendor may be the easy choice, but I believe that hedging the bet might be a better strategy in many situations.

Yuval Boger is a quantum computing executive. Known as the original “Qubit Guy,” he most recently served as Chief Marketing Officer for Classiq.

September 12, 2022