Before the Qubit, the Building

The physical bet behind every quantum headline

Every quantum-computing headline focuses on the machine: more qubits, better fidelity, faster error correction, a new path to commercial advantage. But before any of that becomes useful, someone has to build the physical environment capable of supporting it. A quantum campus is not a data center with a more exotic computer inside. It is a different infrastructure problem.

There is no standard quantum campus

"Quantum computing" describes several competing architectures with very different physical requirements. Superconducting systems run inside dilution refrigerators near absolute zero. Trapped-ion and neutral-atom systems depend on vacuum chambers, lasers, and electromagnetic stability. Photonic systems run warmer, but utility-scale designs can still require industrial cryogenic plants.¹

Each modality demands a different facility — different cooling, different vibration tolerance, different maintenance, different supply chain.

Operation is different too. A data center is legible in real time: pull a drive, read a temperature log, isolate a fault. A quantum processor sealed inside cryogenic or vacuum systems can't be inspected that way. Its performance has to be read through indirect signals, which means instrumentation must be designed in before the machine is ever turned on — not added later.

There is no universal template. There are competing architectures whose eventual scale and commercial relevance remain unsettled.

Power is only one variable

The data-center industry thinks in megawatts — global data-center electricity demand is projected to reach roughly 945 TWh by 2030, more than double 2024 levels.² For quantum infrastructure, power is one constraint among several: ultra-low temperatures, vacuum integrity, vibration isolation, optical stability, electromagnetic interference, and integration with substantial classical computing. NIST is already developing cryogenic electronics and calibration methods that may be required for future million-qubit systems³ — evidence that scaling quantum computing is a systems problem, not just a processor problem.

A conventional facility model asks how many megawatts a campus needs. A quantum model has to ask what operating conditions must stay stable for the system to compute anything useful at all.

The capital risk begins before the technology is settled

Land is being acquired. Facilities are being designed. Governments are funding regional ecosystems — before the winning architecture is known.

IBM is targeting 2029 for its Starling fault-tolerant system at a new Poughkeepsie quantum data center.⁴ PsiQuantum has broken ground in Chicago and Australia.⁵ DARPA's Quantum Benchmarking Initiative is still evaluating whether any current approach reaches industrial usefulness by 2033.⁶ Australia committed roughly US$620 million to PsiQuantum's photonic approach⁷ — a project that has already relocated once, from Brisbane Airport to Moreton Bay, before breaking ground.⁸

None of that makes the investment misguided. It reveals the problem: buildings, public funding, and supply chains are being committed before the dominant architecture exists. A facility optimized for one modality may not support another. A site chosen for power may lack the workforce, partnerships, or specialized utilities the system ultimately needs.

The uncertainty isn't a reason to wait. It's a reason to invest differently.

Stop planning around one future

Most infrastructure is still built around a base case: forecast demand, pick a technology, size the capacity, approve the capital. Quantum isn't moving along one curve. Change the architecture and you change the cooling model, the facility, the site, and the grid and workforce requirements behind it — a chain, not an independent variable.

Before capital moves, the real questions are which investments hold up across multiple technical paths, what should be built now versus left modular, and which decisions permanently lock the project onto one path. These are capital-allocation questions. A single forecast can't answer them.

Measure the useful outcome

The same logic applies to environmental performance. Comparing quantum and classical systems by total electricity consumed won't be enough — quantum infrastructure carries real overhead from cooling, control electronics, and error correction, even when it isn't producing a useful result. A 2026 full-system energy model shows that estimates scoped only to the processor materially understate the real footprint once shared maintenance and classical-processing costs are included.⁹ A separate study models the physical resource cost directly, down to the cooling water consumed removing waste heat from the cryogenic plant.¹⁰

The better measure isn't energy consumed. It's energy to useful outcome — how much infrastructure it takes to produce a result a classical system couldn't reach economically.

The physical bet comes first

The quantum conversation is dominated by qubits and algorithms. But the shift from lab experiment to useful infrastructure depends on something bigger: whether we can build physical environments for a technology still being invented.

The industry keeps asking how to power the next generation of computing. The harder question is how to commit billions to a physical asset when the machine, its environment, and its path to scale are all still unsettled.

No forecast answers that. The decision has to be rehearsed before the campus is built.

---

References

1. PsiQuantum and Linde Engineering, cryogenic-plant partnership for PsiQuantum's Australian facility — PsiQuantum

2. IEA, Energy and AI — global data-center electricity demand projected to reach ~945 TWh by 2030 — IEA

3. NIST, cryogenic electronics and RF-calibration methods for scaling quantum systems — NIST

4. IBM, Quantum Starling fault-tolerant system and Poughkeepsie quantum data center — IBM

5. PsiQuantum, groundbreaking at the Illinois Quantum and Microelectronics Park — PsiQuantum

6. DARPA, Quantum Benchmarking Initiative, evaluating industrially useful quantum computing by 2033 — DARPA

7. Australian Commonwealth and Queensland funding package for PsiQuantum, ~A$940 million — Australian Financial Review; PsiQuantum

8. PsiQuantum, relocation to Moreton Bay and June 2026 groundbreaking — PsiQuantum

9. A Full-System Energy Model for Quantum Computing in an HPC Context, arXiv:2605.09580

10. McCollum et al., "Uncertain Quantum Computing Futures and Potential Energy and Physical Resource Impacts at Scale," Renewable and Sustainable Energy Transition, Vol. 9, Article 100140, 2026. DOI: 10.1016/j.rset.2026.100140