# Can Silicon Photonics Solve Quantum Computing's Scaling Crisis?

Silicon photonics is the connective tissue that quantum hardware needs — and the UK's £2.5 billion Quantum Strategy is explicitly betting on it. That is the core argument advanced by Professor Callum Littlejohns, Deputy Director at CORNERSTONE, and Dr Amit Agrawal, Associate Professor in Optical Engineering at the University of Cambridge, in a technical essay published August 18, 2026.

Their thesis is pointed: no single qubit modality — not [trapped ion](https://quantumintel.tech/glossary/neutral-atom-qubit), not neutral atom, not photonic, not spin — can scale without solving its optical control problem. As qubit counts climb into the hundreds and eventually thousands, the free-space optical setups that work in research labs (manual mirrors, lenses, waveplates, beam splitters) become unmanageable — vibration-sensitive, drift-prone, and impossible to align at volume. Silicon photonics, which inherits decades of CMOS manufacturing precision, offers a chip-scale alternative that is, crucially, modality-agnostic.

The academic authors do not represent a single company's pitch. They represent a fabrication infrastructure argument: the bottleneck is not qubit physics, it is the optical interconnect layer, and the semiconductor industry already knows how to mass-produce it.

---

## Why Optical Control Is the Unsolved Scaling Problem

Every major qubit platform depends on laser-based control. Trapped ion systems use electromagnetic fields to confine ions, then laser beams to manipulate them. [Neutral atom](https://quantumintel.tech/glossary/neutral-atom-qubit) platforms use optical tweezers to hold and address atoms. Spin-based systems — including semiconductor quantum dots and colour centres in silicon and diamond — rely on optical interfaces for initialisation and readout. [Photonic qubits](https://quantumintel.tech/glossary/photonic-qubit) encode information directly in photon states.

In all four cases, the optical delivery requirement is exacting: precise spatial alignment, wavelength selectivity, and phase stability. Littlejohns and Agrawal highlight a specific example: fast, site-selective addressing of large neutral atom arrays demands beams switchable on microsecond timescales. Free-space beam steering cannot reliably sustain that at scale. Integrated photonic switching can.

[Coherence time](https://quantumintel.tech/glossary/coherence-time) preservation — protecting delicate quantum states from environmental disturbance — compounds the challenge. Every additional optical component introduced into a free-space setup is a new vibration coupling point, a new thermal drift source, a new alignment failure mode. Sub-percent variations in waveguide dimensions, the authors note, can affect beam power and phase and, in turn, qubit [gate fidelity](https://quantumintel.tech/glossary/gate-fidelity). The fabrication tolerances required are unforgiving.

Silicon photonics addresses this by integrating the entire optical control layer — generation, routing, modulation, readout — onto a single chip manufactured to semiconductor-grade tolerances. The result is reproducibility at volume, which is the prerequisite for any serious path toward [fault-tolerant quantum computing](https://quantumintel.tech/glossary/fault-tolerant-quantum-computing).

---

## CMOS Compatibility as a Strategic Asset

The authors make a structural argument that deserves scrutiny from investors evaluating quantum hardware plays: silicon photonics is not a quantum-native technology. It is a mature semiconductor technology being redirected toward a quantum application. That distinction matters.

CMOS-compatible silicon photonic processes have been refined over decades for telecoms and data centre interconnects. Foundries capable of producing these chips at high volume already exist. Multi-wavelength operation via wavelength division multiplexing — allowing a single device to address many qubits simultaneously without additional free-space optics — is already demonstrated in classical photonics applications.

This is precisely why [PsiQuantum](https://quantumintel.tech/companies/psiquantum) has staked its architecture on silicon photonics: the company is leveraging standard semiconductor foundries to manufacture what it describes as a path toward fault-tolerant, million-qubit quantum computers, according to the source. Photonic Inc. is cited alongside PsiQuantum as another firm developing optically linked spin qubits in silicon — a notable data point because it confirms that integrated photonics matters even where the qubit itself is not photonic.

The authors' underlying point is that industry has already placed its bets. The academic question is whether fabrication infrastructure — access to advanced fabs, rapid prototyping capabilities, and coordinated research platforms — can keep pace with hardware ambition. They note explicitly that this access varies between countries and may have a significant bearing on the success of national quantum strategies.

---

## The UK's £2.5 Billion Quantum Strategy and Infrastructure Gaps

The UK government's £2.5 billion Quantum Strategy, described by the authors as a 10-year plan targeting the world's first scaled quantum computers, provides the policy backdrop. CORNERSTONE — the co-author's institution — is a UK photonics foundry facility designed to provide exactly the kind of rapid prototyping and fabrication access the authors identify as strategically critical.

The infrastructure argument is the most commercially actionable part of this essay. Quantum hardware companies cannot iterate fast enough on qubit physics if the photonic integration layer requires months-long foundry queue times. Countries and institutions that solve fabrication access — rapid turnaround, CMOS-compatible processes, tight tolerances — will have a structural advantage in the race toward scaled systems.

This also has direct implications for enterprise buyers and investors evaluating quantum platforms. A quantum computing company's roadmap to hundreds or thousands of qubits is only as credible as its optical control architecture. A system whose scaling path depends on manual free-space optics has a hard ceiling. One built on integrated silicon photonics has, at least in principle, a manufacturing-backed scaling path.

---

## Skeptical Notes: What the Essay Does Not Address

The Littlejohns-Agrawal argument is technically sound but stops short of engaging with the genuine unsolved problems in photonic integration for quantum systems.

**Loss**: Silicon photonic waveguides and switches introduce insertion loss. In classical telecoms, loss is tolerable. In quantum systems, where single-photon detection and high-fidelity state readout are required, loss directly degrades [gate fidelity](https://quantumintel.tech/glossary/gate-fidelity) and measurement outcomes. The source does not quantify what loss levels current silicon photonic platforms achieve in quantum-relevant configurations.

**Cryogenic compatibility**: Many qubit platforms — superconducting transmons, in particular — operate at millikelvin temperatures. Silicon photonic components are typically designed for room-temperature operation. The source does not address whether and how silicon photonic integration functions at cryogenic temperatures, which is an active and unresolved research area.

**Crosstalk**: Packing dense photonic switching networks onto a chip in close proximity to sensitive qubits creates electromagnetic and thermal crosstalk risks. This is not mentioned.

These are not reasons to dismiss the silicon photonics thesis — they are reasons to hold platform vendors accountable for specific integration benchmarks before accepting scaling roadmaps at face value.

---

## Industry Trajectory Implications

The guest post reflects a maturing consensus: the quantum industry's bottleneck is shifting from qubit physics to engineering integration. The companies and national programs that will lead the next phase of quantum computing development are those solving the optical, cryogenic, and packaging engineering problems — not simply adding more physical qubits.

For quantum engineers evaluating architectural decisions: the CMOS foundry compatibility argument is the most durable part of the silicon photonics case. Any quantum system that requires custom optical components assembled by hand cannot be manufactured at scale. For Series A and B investors evaluating quantum hardware companies: ask specifically how the optical control layer scales with qubit count, what the fabrication partner is, and what loss and crosstalk specifications the system achieves. For enterprise buyers: qubit count is a lagging indicator. The optical integration architecture is a better early signal of whether a vendor's scaling roadmap is engineering-grounded or aspirational.

---

## Key Takeaways

- **Silicon photonics is modality-agnostic**: it can serve trapped ion, neutral atom, spin, and photonic qubit platforms by solving the shared optical control scaling problem.
- **The UK's £2.5 billion Quantum Strategy** explicitly targets scaled quantum computers; fabrication infrastructure access is identified as a key variable in national competitiveness.
- **PsiQuantum and Photonic Inc.** are cited as companies already betting on silicon photonic integration — one for photonic qubits, one for spin qubits.
- **CMOS manufacturing heritage** is the key argument: decades of semiconductor precision manufacturing can be redirected to quantum optical layer production.
- **Waveguide fabrication tolerance** is critical: sub-percent dimensional variations can affect gate fidelity, making fabrication-aware design non-optional.
- **Unresolved challenges** — optical loss in quantum configurations, cryogenic compatibility, and crosstalk — are not addressed in the source and must be tracked independently.
- **Fabrication access** (rapid prototyping, advanced fabs, coordinated research platforms) varies by country and may determine which national quantum programs succeed.

---

## Frequently Asked Questions

**What is silicon photonics and why does it matter for quantum computing?**
Silicon photonics uses CMOS-compatible semiconductor manufacturing to build chip-scale optical circuits that generate, route, modulate, and detect light. In quantum computing, it matters because every major qubit modality — trapped ion, neutral atom, photonic, and spin — requires precise optical control to initialise, manipulate, and read out qubits. As qubit counts rise, free-space optics become unmanageable; integrated silicon photonics provides a manufacturable, scalable alternative.

**Which quantum computing companies are using silicon photonics?**
[PsiQuantum](https://quantumintel.tech/companies/psiquantum) is the most prominent, using standard semiconductor foundries to pursue fault-tolerant quantum computing via photonic qubits. Photonic Inc. is developing optically linked spin qubits in silicon, demonstrating that silicon photonics integration matters even when the qubit itself is not photonic. Both are cited in the August 2026 analysis by CORNERSTONE and University of Cambridge researchers.

**What is the UK's Quantum Strategy and how much has been committed?**
The UK government has committed £2.5 billion over 10 years through its Quantum Strategy, targeting deployment of scaled quantum computers. The program treats fabrication infrastructure — including access to photonics foundries — as a strategic priority.

**Why can't quantum systems just use free-space optics at scale?**
Free-space optical setups (mirrors, lenses, waveplates, beam splitters assembled manually) work in research labs but become unreliable as qubit counts grow. They are sensitive to vibration, prone to thermal drift, and require manual alignment that is time-consuming and not reproducible at volume. For neutral atom platforms, site-selective addressing on microsecond timescales is particularly difficult with free-space steering.

**What are the main unsolved problems in silicon photonics for quantum systems?**
Three key challenges not resolved in current literature: (1) optical insertion loss, which degrades single-photon detection fidelity; (2) cryogenic compatibility — many qubit platforms operate at millikelvin temperatures where standard silicon photonic designs were not validated; (3) electromagnetic and thermal crosstalk between dense photonic switching networks and proximate qubits. These represent the next engineering frontier for the silicon photonics-quantum integration agenda.