Hardware Modality Diligence
Superconducting, trapped-ion, neutral-atom, photonic, silicon-spin — every hardware pitch opens with a modality, and every modality sounds inevitable from the inside. This chapter gives you a way to compare them: a matrix of physical facts, weak layers, and proof gates that turns a hardware pitch into a scoreable memo.
A quantum hardware company is two bets wearing one logo: a bet that its qubit modality can scale, and a bet that this particular team can execute the scaling. Most diligence scores the first bet and assumes the second.
By the end of this chapter you will be able to build a modality comparison matrix for any three hardware companies, fill it with evidence instead of adjectives, and name the layer of each stack where the thesis is most likely to die.
Core concepts: hardware modalities, modality proof gates, company diligence.
The modality question is two questions
Hardware diligence separates modality promise from company execution. Superconducting circuits, trapped ions, neutral atoms, photons, silicon spins, bosonic encodings, and topological approaches each carry real strengths and real bottlenecks, and each has produced confident predictions of victory. The modality debate is genuinely interesting physics. It is also mostly the wrong question for a memo.
The better question is: which proof gate would change your confidence for this modality, this company, and this workload? A proof gate is a measurable event that forces an update — gate quality held while the device scales, a demonstrated logical building block, a readout target hit, modules linked into a larger system. If you cannot name the next proof gate, you do not have a thesis yet. You have a preference.
A scorecard you can defend
Score each company's claims with four terms:
- Claim quality. Is the claim specific, measured, relevant to the workload, and labeled with a source?
- Proof progress. What evidence exists against the next proof gate?
- Risk. Technical, capital, supply, and timing exposure.
- Kill-criteria pressure. How close is the observed evidence to the conditions that would end the thesis?
A hardware claim earns a higher score only when its evidence is specific, measured, relevant, and source-labeled — never when it is simply louder.
The second tool is the logical–physical estimate: . It stops the memo from treating raw physical qubit counts as the final scaling metric. What matters is logical operations at a target error rate, and the overhead needed to get there. A company reporting a large physical array has told you its input to this equation, not its output.
Worked example: three companies, one matrix
Take three companies using different modalities and put them in a single matrix with the same rows: physical qubit count, gate mechanism, readout, native connectivity, dominant errors, calibration burden, infrastructure demands, error-correction code path, logical-operation evidence, and next proof gate. Shared rows are what make the comparison honest — each company must answer every question, including the ones its marketing avoids.
The matrix does not declare a universal winner. It makes the bets visible. Every company carries a modality thesis (this physics scales) and an execution thesis (this team gets there first), and you score the two separately. A strong modality with a weak team is a licensing story. A strong team on a weak modality is a pivot waiting to happen.
For each cell that matters, write the next proof gate in one sentence — quality maintained at larger scale, a logical building block, improved readout, modular links — and write next to it what would count as failure.
Where the memo goes wrong
The first trap is choosing a favorite modality before specifying workload and scale. A high-fidelity small system, a large low-fidelity array, and a promising component do not answer the same question, and ranking them against each other without a workload produces a ranking of your biases.
The second trap is overvaluing the metric the company emphasizes while ignoring the layer it mentions least. A useful matrix names what would make the thesis worse: quality that falls as scale rises, readout that cannot support correction, packaging that blocks control density. Any of those should lower confidence even while a headline metric improves. The weakest layer must be visible in the memo before any ranking appears.
The third trap is treating a company roadmap as neutral technical evidence. A roadmap is a claim about the future issued by an interested party. Chapter 81 is devoted to reading them.
The engineering view
For a computer scientist, modality diligence is a cost-model exercise. The effective cost of a circuit depends on connectivity, native gate set, operation duration, error rate, readout, reset, and the compilation pass that maps the algorithm onto all of it. The same algorithm stresses modalities differently — a nearest-neighbor-heavy circuit and a dense all-to-all circuit are different machines' problems.
This is why a modality memo has to connect hardware to software. The compiler, runtime, decoder, and application layer all inherit the modality's constraints. If the company cannot expose reliable operations to the software stack, the modality thesis stays incomplete no matter how strong the physics paper is.
What this buys you in diligence
A partner decision can be justified when a hardware company exposes a useful testbed — access is itself evidence. An investment memo needs more: a scaling path, defensibility, capital requirements, and proof gates that arrive in sequence. Wait or avoid is the right label when claims are broad but the evidence remains narrow.
Educational diligence should never read as a trade instruction. State what technical evidence is missing and what decision you would make in a study setting, and stop there.
Exercise
Compare three modalities. Pick three hardware companies using different approaches and write a one-page modality diligence matrix covering the rows from the worked example.
- Submit: the matrix, one next proof gate per company, and two kill criteria per modality.
- Check: apply the decision score and the logical–physical estimate to each company. Underline every number in your matrix and write its source beside it; any number without a source gets relabeled as a claim.
- Decide: for one hardware category, choose partner, monitor, wait, or avoid — and write the evidence that would change the label.
- Repair: if your matrix ranks companies by raw qubit count or brand familiarity, redo it after Chapter 73.
Check your understanding
Without notes: write a modality diligence matrix comparing three hardware approaches, with strengths, bottlenecks, and kill criteria for each.
A passing matrix separates the modality thesis from the execution thesis, names the weakest layer of each stack, and gives every metric a source label.
Oral defense: take one company's public roadmap and argue, in two minutes, which single milestone would most change your confidence — and which milestone is doing no evidentiary work at all.
If you get stuck
If your matrix repeats roadmap language without source labels, skips missing evidence, or omits kill criteria, revisit Chapter 73 (The Quantum Company Landscape) for the company-level map and Chapter 54 (The Full Quantum Computer Stack) for the layers your matrix rows come from.