Guest post by Professor Callum Littlejohns, Deputy Director at CORNERSTONE, and Dr Amit Agrawal, Associate Professor in Optical Engineering at the University of Cambridge
The ability to scale is one of the most pressing issues in quantum hardware today, and solving scaling challenges has become a driving force behind significant government investments. This includes the UK government’s £2.5 billion Quantum Strategy – a 10-year plan for deploying the world’s first scaled quantum computers.
But to understand how to achieve scaled quantum systems, we need to look outside of the quantum discipline, to other tried-and-tested technologies.
In many ways, the quantum scaling challenge mirrors those faced by other industries before it: larger systems require more hardware, which means more complexity. However, quantum systems differ in that system stability relies on protecting delicate quantum states from the slightest disturbances to maintain coherence, increasing the complexity of the challenge. The hardware is constantly evolving, and as you add more components, this isolation is harder to maintain. Not to mention that you quickly run into other problems like added heat, latency, alignment complexity, and packaging challenges.
In 2026, silicon photonics is emerging as a leading candidate for building scalable quantum systems. It provides integrated, chip-scale circuits able to generate, manipulate, route and read out qubits and deliver light to, and potentially collect light from, qubits realised in separate physical modalities, e.g. photonic, spin, trapped ions or neutral atoms. In both cases, decades of evidence in the CMOS industry show that we can manufacture integrated silicon circuits at huge volumes with extreme precision, which is critical for scalability.
To maintain momentum, the sector’s overall progress – and its ability to support quantum scaling – hinges on infrastructure, including access to advanced fabrication, rapid prototyping, and coordinated research platforms. This access varies between countries and may have a significant bearing on the success of national quantum strategies.
Silicon photonics as the quantum optical layer
Silicon photonics is a qubit modality-agnostic solution with the ability to support quantum computing platforms relying on trapped ions, neutral atoms, and photonic qubits, since each depends on highly precise optical control. Trapped ion systems confine individual ions using electromagnetic fields, neutral atom platforms hold and manipulate atoms with optical tweezers, while photonic approaches encode information directly in the quantum states of photons themselves. Spin-based platforms, from semiconductor quantum dots to colour centres in silicon and diamond, likewise rely on precise optical interfaces to initialise, read out, and interconnect qubits. In every case, multiple laser beams must be delivered with exact spatial alignment, wavelength selection, and stability to manipulate qubits reliably. Fast, site-selective addressing of large neutral atom arrays, for example, demands beams that can be switched and modulated on microsecond timescales – difficult to sustain with free-space beam steering, and a natural role for integrated photonic switching. Efficient on-chip collection of the light emitted by these qubits is also being actively explored, a capability that would underpin both high-fidelity readout and photonic links between processors. As qubit counts climb, this becomes even more important.
Silicon photonics is a natural fit for pre-existing semiconductor manufacturing, with CMOS-compatible processes, and encouragingly, industry players are already placing bets on this technology. For example, PsiQuantum is leveraging silicon photonics to manufacture fault-tolerant, million-qubit quantum computers using standard semiconductor foundries. Photonic Inc., meanwhile, is developing optically linked spin qubits in silicon, a reminder that integrated photonics matters even where the qubit itself is not photonic.
As with other deep technologies, quantum computers first came to life in the lab, relying on manual assembly and calibration. Free-space optical setups – manually assembled mirrors, lenses, waveplates and beam splitters – can achieve the required control in smaller systems, but they quickly become unwieldy as qubit counts rise into the hundreds or thousands: aligning multiple beams by hand is time-consuming, sensitive to vibration, and prone to drift. But as the technology matures from scientific demonstrations to commercial engineering, the setup is evolving. By integrating complex optical components onto a single chip, silicon photonics supports this evolution by delivering precise, reproducible quantum circuits in a compact form. Multi-wavelength operation can also be realised on-chip through wavelength division multiplexing, allowing a single device to address many qubits without additional free-space optics.
Ultimately, the route to commercialisation hinges on this shift towards integration. Because each integrated device is produced to exceptionally tight tolerances, variability between chips drops dramatically, creating the predictability and quality that will enable high-fidelity quantum operations at scale. The margins are unforgiving: even sub-percent variations in waveguide dimensions can affect beam power and phase – and, in turn, qubit fidelity – making fabrication-aware design, where layouts compensate for predictable process deviations, essential.
Silicon photonics is already a mature, foundational technology within data centre and AI infrastructure, where it is used to efficiently move massive amounts of data. This is advantageous for quantum, since there is a pre-existing manufacturing ecosystem available to leverage. This commercial maturity is what makes silicon photonics a viable candidate for a mass-manufacturable quantum computer.
Connecting hybrid, modular systems
Another factor making silicon photonics so promising for quantum is the shift towards hybrid, modular architectures. In hybrid systems, quantum processors are used for specific simulations or optimisation tasks, while classical processors handle pre-processing, quantum error correction, and post-processing.
Silicon photonics is well placed to serve as the interconnect layer between these parts. Optical links can move data between the quantum core and classical control units at high bandwidth and low latency, without the thermal load of electrical interconnects. Modular designs also allow multiple photonic chips to interconnect, laying the foundation for rack-level systems and, eventually, distributed quantum networks.
The long-term vision for quantum computing is commercial-grade systems that can be deployed in real-world environments. Silicon photonics is one of the most promising technologies we have that can provide the combination of density, stability, and compatibility with existing semiconductor manufacturing. But hardware compatibility alone isn’t enough. The speed at which teams can iterate is what will separate them.
Emerging technologies are notoriously competitive, so having access to the right infrastructure, which allows researchers and industry R&D teams to prototype quickly and produce updated versions, creates a better chance of overcoming early-stage hurdles.
Multi-project wafer runs, for example, allow several designs to share the same wafer, lowering costs and speeding up experimentation. This model is already widely employed by foundries in the semiconductor manufacturing ecosystem. But this is only one piece of the puzzle.
The emphasis on developing sovereign tech capabilities is growing stronger for governments around the globe, making access to scaling infrastructure critical. This is particularly important for silicon photonics due to the myriad of applications it will be the enabling platform for. Given current regional disparities, there is a risk that the nations without the infrastructure in place will fall behind.
Fortunately, the tide is turning, and there is a growing recognition from governments of the need to build pilot lines to bridge the gap between research and manufacturable solutions. Last year, for example, the EU announced PIXEurope, a €400M initiative to establish a dedicated pilot line for photonic integrated circuits. In the UK, the government has committed up to £2 billion for quantum technologies, with £90 million earmarked for infrastructure, though a dedicated photonics pilot line remains in discussion.
Silicon photonics is not the whole answer to quantum scaling, but it is increasingly a necessary part of the answer. If the field succeeds, it will not be because quantum systems got larger on their own. It will be because the hardware architecture around them became scalable.
