Quantum computing often gets marketed like a chipbuilding story.
The process is proposed to be a tidy sequence of performing basic science, developing and then miniaturizing components, scaling them, and then repeating the blueprint over and over again. It's an attractive myth because it feels familiar, manageable to understand, and most of all because it fits a narrative that implies the inevitability of cost curves.
The reality is closer to quantum computers being a scientific apparatus that happens to be able to perform certain computations. When the conditions are just right, it can even perform those computations somewhat accurately, at least for a while.
So without further ado, let's dive in and explore why it's so hard to build a quantum computer.
A quantum computer is a finicky scientific apparatus
Today, a quantum computer is still a lab-scale machine. There's a whole industry dedicated to growing them into being a usable technological platform commodity like classical computers are, but they aren't quite there yet.
The elephant in the room is that quantum computers don't behave like a commodity computer, and they might not ever. Instead, they behave more like a delicate experiment that you keep stable just long enough to get the job done. Then it's back to drydock to spend some more effort and money keeping it alive and making sure it's ready to go for the next run. These devices exist at the furthest part of the bleeding edge of humanity's technological capabilities, so seamlessness is not to be expected anywhere.
This is essentially why it's so hard to build a quantum computer today. You're effectively assembling a stack where refrigeration, wiring, fabrication, calibration, and software have to work in tune, at the same time, and for long enough to matter from the perspective of a user. It's the last part which is the most common stumbling block as of now.
Before we go deeper, let's surface the underlying constraints here for anyone seeking to build and/or operate a quantum machine:
Can you reliably fabricate superconducting devices at yield?
Can you operate ultra-low-temperature infrastructure continuously?
Can you route and stabilize a dense set of control and readout lines?
Can you keep calibration stable for long runs?
Can your software compile around hardware constraints?
For the median aspiring quantum computer builder or operator today, the answer to most of those questions is either a simple “no” or, at best, a “not consistently.”
That makes sense when you zoom in on what modern superconducting qubits literally are. They're patterned circuits with nonlinear elements called Josephson junctions. This is device physics, laid out on a chip, and then coaxed into behaving like an artificial atom.
That element immediately drags in the issue of manufacturing and repeatability; IBM has argued that getting to scale demands advanced workflows, with the company opting to use 300mm semiconductor fabrication rather than artisanal one-off processing. Yield, uniformity, and metrology are now as important as clever algorithms, but no single element of these processes are significantly refined yet.
That's before even getting into the software layer, which is where all the hardware weirdness becomes visible by necessity. Nonetheless, software is advancing quickly, and it's already ahead of where the hardware can go. For instance, IBM's Qiskit is an open-source SDK for building and executing circuits on real devices.
There's an important security implication here; if building a quantum computer means building a lab-grade stack with specialized fabrication, calibration, and runtime software, a lone actor in a basement is structurally disadvantaged. That doesn't eliminate the risk of such an attacker emerging in the future, but it does shift the default threat model toward well-funded organizations, like governments.
Let's take a closer look at the required physical infrastructure behind quantum computing to clarify why that's the case.
Cryogenics and shielding are the real factory floor
It is tempting to say that quantum computers have to be cold and stop there, but that's underselling it. The more accurate statement is that the machine’s physical environment is an inseparable part of its ability to perform computations, and controlling that environment to the point where computations are reliable is a major engineering project from the get go.
Dilution refrigeration is one of the defining pieces of hardware for many platforms. The objective of this refrigeration method is to keep the refrigerated components cooled to a range of typically 10 to 20 millikelvin, which is hundreds of times colder than the depths of outer space. To accomplish that, a dilution refrigerator cools things by forcing helium-3 atoms to mix into helium-4, a process that naturally absorbs heat. By continuously pumping helium-3 across this boundary, the system pulls thermal energy out of whatever it’s attached to, reaching near-absolute-zero temperatures.
But before that can even happen, the entire apparatus needs to be cooled to a temperature of several kelvin, so it's first pre-cooled using conventional cryogenics, typically via a bath of liquid nitrogen followed by a bath of liquid helium or with closed-cycle cryocoolers. The key point for a practical reader is that this process requires a parade of specialized infrastructure working in sequence, with its own heat budgets, failure modes, and maintenance requirements.
Once the core hardware is kept cold, you then have a new enemy; the heat carried by everything connected to it. One of many practical challenges is signal transfer across temperature stages.
Every added line is a thermal and mechanical liability, and the liability gets even worse as you scale. Think about the computer as a 3D object composed of many 3D components, not as a flat chip. Understand that the verticality and minimum viable shape of key components implies heat transferring through different media, typically at very different rates, all of which must be accounted for in complex ways.
Furthermore, note that heat transfer to the air can introduce other problems that need to be guarded against, like condensation. At the low temperatures involved, every element in the air, like nitrogen, oxygen, argon, and CO₂ become cold enough to condense first into a liquid, and then frozen into a solid, so the entire operation has to take place in a vacuum container.
Maintaining that vacuum state is also quite technologically burdensome, and of course the vacuum needs to be safely equalized with the ambient air before it's possible for staff to make any physical changes to the hardware, and then re-established afterward. The table below outlines recurring issues that make a quantum computer expensive, slow to scale, and hard to replicate outside industrial settings.
Constraint | Why it matters | What tends to scale poorly |
|---|---|---|
Ultra-low temperatures and stable thermal stages can be central to qubit stability | Cooling capacity, uptime, and maintenance complexity | |
Each added line and component consumes limited cooling power | Cable count, routing density, and integration effort | |
Scaling input-output hardware is a bottleneck for fault-tolerant systems | Thermal modeling, verification, and hardware design complexity | |
External energy sources can cause multi-qubit correlated errors, undercutting key error assumptions | Shielding strategies and tail-risk mitigation |
Ultra-low temperatures and stable thermal stages can be central to qubit stability
Cooling capacity, uptime, and maintenance complexity
Each added line and component consumes limited cooling power
Cable count, routing density, and integration effort
Scaling input-output hardware is a bottleneck for fault-tolerant systems
Thermal modeling, verification, and hardware design complexity
External energy sources can cause multi-qubit correlated errors, undercutting key error assumptions
Shielding strategies and tail-risk mitigation
The thing to remember here is that a quantum computer lives or dies on infrastructure.
Now we can connect the physical constraints to the computational constraints.
Error correction makes reliability non-negotiable
The reason quantum computers are hard to make is that today’s qubits are noisy, and scalable computation depends on suppressing that noise by improving the performance of the various apparatuses we discussed above. Nature’s report on a recent milestone underlines that error correction works by encoding logical qubits from physical qubits, and, critically, it's the logical error rate that ultimately matters for the purpose of making computations.
That has a non-obvious consequence; if error rates are not comfortably low, the system spends an enormous fraction of its hardware budget on redundancy, and pretty much every element of the system is already quite expensive as well as physically large (at least compared to similar components in classical computers). A quantum computer can thus grow tremendously in physical size while barely growing at all in useful capability.
Operationally, error correction also makes drift expensive. Calibration repetition is a tool for managing parameter drift, and as devices scale, calibration itself becomes an engineering discipline. There's thus the need for frequent calibration, because drift undermines long computations and ultimately makes the most useful computations the most difficult to perform reliably. So you can't just assemble the machine once, it needs to be tuned, and the tuning cost grows with the system.
Once you accept that keeping a quantum computer functional is ongoing and fairly capital intensive work, it's possible to start thinking about who can afford to do that work at scale.
Supply chains and talent keep quantum industrial
Scaling a quantum computer means buying, operating, and integrating specialized subsystems that are not yet commoditized. That reality shows up at the materials layer and at the labor layer, because supply chains and expertise determine who can build and maintain large fleets.
Multiple chokepoints, including helium-3 for dilution refrigeration, can slow deployment of quantum infrastructure and has the effect of concentrating capability into a small number of actors.
On the other hand, not every approach to quantum computing requires the same type of deep expertise in cryogenics. Neutral-atom systems can shift the burden toward optics and vacuum engineering, with neutral-atom quantum computers built around a vacuum chamber with laser-based atomic cooling and optical tweezers. That's obviously still not garage tech, it's simply a different flavor of extremely specialized and finicky infrastructure.
The path to quantum computers being broadly available thus looks more like building an advanced scientific-instrument industry than like shipping a new smartphone. In closing, to find investment insights for the quantum computing supply chain, the best move is thus to look for what constrains scaling.
To keep up with the latest in blockchain technology and quantum computing, join us on X and subscribe to our newsletter.
Sources
Building quantum computers with leading-edge semiconductor fab
Engineering cryogenic setups for 100-qubit scale superconducting circuit-based quantum computing
Thermal Capacity Mapping of Cryogenic Platforms for Quantum Computing
Cosmic-ray-induced correlated errors in superconducting qubits
QatarEnergy signs long-term helium supply deal with Germany's Messer
Fast-feedback protocols for calibration and drift control in quantum computing
In-situ Qubit Calibration for Surface Code Quantum Error Correction
Quantum Computer Controlled by Superconducting Digital Control Electronics
