IQM's "Razor & Blades" Dragon
An Informal Case Study in Growing Market Share
Are you familiar with the “Razor & Blades Business Model” of Gillette (Procter & Gamble)? Gillette grabbed market share by offering an inexpensive handle for safety razors. The minimal initial investment eased adoption.
This stuck with me when reading IQM’s recent announcement about opening an office in Seoul, South Korea, its second Asia-Pacific office, after Singapore. The announcement also mentioned the sale and four-month deployment of a five-qubit IQM Spark to Chungbuk National University (CBNU). IQM Spark costs less than EUR 1M, which means that its relatively minimal initial investment eases adoption.
Time to Upgrade
Once you bought a Gillette razor handle, you were locked into Gillette’s consumable razor blades. Not only were you buying replacement blades, you were probably upgrading at some point to more-expensive multiple-blade cartridges. You might buy systems with novel features, and you might even buy higher-end handles. I remember dry-shaving with single-blade razors as a teenager, but I use five-blade badboys today.
Now imagine you’re the proud owner of an IQM Spark, and it’s time to upgrade beyond five qubits. You know IQM, you like IQM, you’re satisfied with their support, and you’ve already trained your workforce. If IQM Spark is analogous to a single-blade inexpensive razor, then IQM Radiance is a multiple-blade higher-end razor. The business model worked for Gillette, and it makes sense here, too.
Where in the world is IQM?
Counting over 300 employees, IQM has its headquarters in Finland and has a presence in Germany, France, Italy, Poland, Spain, Singapore, South Korea and the United States. Now where you see the word “presence,” I see “market share.” And with an installation ongoing in Taiwan, IQM is expanding beyond these ten countries. I also happen to know that IQM Spark has been discussed in the Philippines.
I haven’t done the specific research, but I am unaware of any other quantum computing company with such a global presence. After all, when your NISQ devices cost tens of millions of dollars, you can’t expect to sell too many of them. Your geographic focus narrows to where such budgets can be found. But IQM Spark is affordable, in comparison, and look at IQM spreading across the world map.
Getting Started
What do you get for under EUR 1M? I’m not talking about qubits anymore. Compared to spending tens of millions of dollars, you get a simpler, shorter acquisition process. But wait, there’s more! Instead of waiting one or two years for your deployment, you get a lead time as short as six months (if IQM focuses on Spark). If you’re feeling a little behind in the quantum race, this might be the quickest way to get on board. You’re rewarded with an open hardware and software platform that supports popular SDKs, as well as pulse control and HPC integration.
IQM sells so many systems that it actually has an assembly line. It has already built 30 systems, with an unspecified split between Spark and Radiance. The company could conceivably build 20 Sparks a year and deliver them with a 6-month lead time if it were to focus only on Sparks; it can deliver almost every other system in 6-12 months. The first sale was in May 2020; systems are currently selling at a rate of one per month, and we recently saw two deliveries on one day. Your choices are the 5-qubit Spark, with pulse control coming later this year, or a 20-, 54-, or 150-qubit square lattice Radiance. One special delivery to the Czech Republic has a 24-qubit star topology, which is my personal favorite.
Conclusion
VTT’s single-blade inexpensive first “razor” was a five-qubit superconducting quantum computer from IQM. It has since bought from IQM multiple-blade, higher-end “razors” in the form of 20-qubit and 50-qubit computers. It was recently announced that VTT will receive even higher end “razors” in the near future: a 150-qubit IQM Radiance in 2026 and a 300-qubit IQM Radiance in 2027.
I suspect that VTT is not going to be a unique case study.
Image generated by Google's language model AI.