You can now buy a DIY quantum computer - New Scientist

Qilimanjaro Quantum Tech is pitching EduQit as a “build-your-own quantum computer,” and for once the fine print is more interesting than the slogan. The Barcelona company says the kit gives universities and research labs a real on-site superconducting system to assemble with help from Qilimanjaro engineers, rather than another cloud login dressed up as “access.” That matters. Quantum education is full of abstractions and dashboards. Actual dilution refrigerators, microwave control electronics and calibration routines are where the field stops being mysticism and starts becoming expensive engineering.

According to Qilimanjaro’s product material and company announcement, EduQit is a modular kit built around a real deployable superconducting system, not a simulator. The company explicitly says it is not intended to maximize performance or prove “quantum advantage.” That disclaimer is doing a lot of honest work. This is a five-qubit machine, not a competitive frontier processor, and Qilimanjaro’s own FAQ frames the product around education, capability building and system-level understanding rather than benchmark supremacy.

That makes the product easier to take seriously than the headline might suggest. In quantum, “you can buy a quantum computer now” is usually a sentence with several concealed trapdoors. Sometimes it means cloud access. Sometimes it means a component, not a system. Sometimes it means a demo unit whose main job is to raise the next round. EduQit appears to be something more concrete but also more limited: a supervised path for institutions that want staff and students to learn how a superconducting machine is assembled, wired, calibrated, operated and modified on premises.

New Scientist reported that the system starts around €1 million, includes training that can run up to three months, and may take at least 10 months to get fully running. Those timelines alone tell you this is not a Raspberry Pi for the chemistry department. It is still a specialized laboratory project with a seven-figure price tag, long integration lead time and obvious infrastructure requirements. “DIY” here means flatpack in the same sense a particle detector might be flatpack: the parts arrive, but so do the constraints.

The more revealing comparison is not to IBM’s or Google’s roadmaps but to other on-premises offerings. Rigetti said when it launched its Novera product in 2023 that the 9-qubit QPU would start at $900,000, but that sale was for the processor unit meant to integrate with third-party cryogenic and control systems, not a full educational stack. Qilimanjaro is trying to sell the stack and the knowledge transfer around it. That shifts the product from “small quantum computer for purchase” to “training platform for building a quantum hardware program.” Those are not the same market.

There is also a strategic subtext. Qilimanjaro says the kit is meant to support workforce development, institutional capability and what it calls technological sovereignty. Strip away the brochure language and the point is straightforward: if universities and regional labs only ever touch quantum hardware through US cloud platforms, they do not build much local know-how. EduQit is a bid to make hands-on quantum operations a teachable institutional capability, not just a remote service consumed from elsewhere.

The technical caveat is that Qilimanjaro’s broader pitch bundles several things together. The company emphasizes its fluxonium-based superconducting approach, multimodal architecture, and analog, digital and hybrid workflows through its software stack. That is interesting, but the educational product should not be confused with proof that its preferred hardware strategy has won anything. A five-qubit teaching system can be useful without saying much about commercial quantum advantage, error-corrected scale, or the timeline to practical applications.

So yes, someone can now buy something that credibly counts as a quantum computer kit. No, this does not mean quantum hardware has become cheap, easy or democratized in any normal sense. What it does suggest is narrower and still meaningful: parts of the superconducting stack are becoming modular enough to package for training, and at least one vendor thinks the next market is not just selling compute time but selling the painful education required to operate the machine in the first place.