Qubit Modalities: Photonic Quantum Computing

Fourth in a series on quantum computing modalities — covering how it works, who the key players are, and where I think photonics is headed.

Introduction to Qubit Series

This is the fourth post in my series on qubit modalities. I’ve already covered Neutral Atoms, Trapped Ions and Superconducting. If you’ve been following along, you know that each modality has its own physics, its own strengths, and its own set of hard engineering problems to solve. Photonics is no different — but it has a few structural advantages that are making very serious investors and governments write very large checks.

In this post, I’ll walk you through how photonic quantum computers work in plain English, what makes them different, who the main players are, and I’ll pick an “Alpha Dog” and a “Breakout Contender”. I’ll also spend a moment on the elephant in the room: PsiQuantum, the most-funded entity in quantum hardware, and what it means for the broader market if their bet doesn’t pay off.

How Photonic Quantum Computers Work (Plain English)

Take a quantum computer and ask yourself: what if the qubits themselves could also carry information across a fiber-optic network? What if the hardware was manufactured in the same semiconductor fabs that make telecom chips? What if you didn’t need a dilution refrigerator the size of a chandelier to run the thing? That’s the photonic quantum computing pitch — and it’s why this modality has quietly become one of the most heavily funded bets in all of quantum. In 2025, photonic quantum hardware attracted more private capital than any other quantum hardware sub-category.

Photonic Quantum Computers, as the name suggests, operate using light. Not laser beams shooting around a room — but single particles of light, called photons, guided through tiny etched channels (waveguides) on a silicon chip, roughly the way electricity flows through wires in a conventional circuit. Each photon can carry quantum information in its properties: its polarization, its phase, or how it is squeezed or entangled with another photon. Those photons are your qubits — but unlike electrons sitting on a chip or atoms floating in a trap, photons are constantly moving. You can’t just park them and wait. You have to run the computation as the light flows through the circuit.

Image Source: https://news.ucsb.edu/2019/019679/pushing-quantum-photonics

A good analogy: think of a photonic quantum processor as a very sophisticated pinball machine for light. Photons enter, hit beam splitters (which act like partial mirrors, sending photons one way or the other probabilistically), pass through phase shifters (which nudge their quantum state), interact, interfere with each other, and are finally detected at the output. The pattern of detections encodes the answer. Because photons barely interact with the environment, they don’t “decohere” the way other qubit types do when exposed to heat or stray fields. That’s a significant natural advantage. The catch: because photons barely interact with anything, including other photons, getting two photons to interact with each other to produce a two-qubit gate is genuinely hard.

There are two main architectural approaches worth knowing:

1. Linear Optical Quantum Computing (LOQC), including the “fusion-based” measurement approach favored by PsiQuantum: qubits are encoded in single photons; gates are performed using beam splitters, phase shifters, ancilla photons, and measurement. The operations are probabilistic — they don’t always succeed — so the systems use massive redundancy and clever resource state engineering to make computation reliable. PsiQuantum’s fusion-based approach has made architectural advances that improve loss tolerance and resource efficiency.

2. Continuous-Variable (CV) quantum computing, including the Gaussian Boson Sampling (GBS) approach pioneered by Xanadu: instead of encoding information in single photons, information is encoded in the amplitude and phase of “squeezed light” — light states where quantum noise has been redistributed. These systems naturally run at room temperature and have a more manageable architecture for certain tasks, but need additional structure (GKP qubits, described below) for full fault tolerance.

Three key technical levers drive performance in photonic quantum computing:

Photon loss is the single biggest technical challenge. Every beam splitter, waveguide, coupling, and connector introduces loss — meaning photons (and the quantum information they carry) simply disappear. Photon loss is such a big hurdle because in photonic quantum computing each photon is like a unique playing card in a very fragile magic trick: if even one card quietly falls out of the deck, the whole trick stops working and you usually cannot even tell exactly where it went wrong. Photon loss is still quite significant, with nearly 30% of photons generated either not detected or lost, so if the loss of a single photon can be catastrophic to a given algorithm, losing 30% is essentially a non-starter. Fault-tolerant photonic computing requires loss thresholds below roughly 0.1–1% per operation so a lot of engineering work still needs to be done.

Chip-scale manufacturing is photonics’ superpower. Photonic chips can be manufactured in existing CMOS semiconductor fabs using established processes. PsiQuantum’s Omega chipset was designed and fabricated on full-size wafers at GlobalFoundries’ facility in New York. This means the cost structure and scalability trajectory could look more like consumer electronics than like bespoke scientific instruments.

Native networking is photonics’ secret weapon for the long game. Photons already are the medium of fiber-optic communication. A photonic quantum computer can, in principle, share quantum information across a network using existing telecom infrastructure. Xanadu’s Aurora demonstrated a modular architecture of four interconnected server racks connected by 13 km of optical fiber — all at room temperature. Photonic Inc. has demonstrated quantum entanglement transfer at telecom wavelengths over point-to-point connections. This built-in networking compatibility is what every other modality has to engineer around but is inherent in photonics’ base architecture.

If you remember only three things about Photonic Quantum Computers, remember these:

1. Room temperature and telecom-compatible. Photons barely interact with thermal noise, and they already travel through fiber-optic cables. This gives photonics a structural cost and networking advantage that no other modality can match natively.

2. Manufacturing at semiconductor scale is the path to a million qubits. By leveraging existing CMOS fabs, photonic systems have a credible roadmap to the qubit counts needed for fault-tolerant computing — something that remains genuinely hard for most other modalities.

3. Photon loss is the Achilles Heel. The physics are beautiful, but engineering is brutal. Every connection is a potential error. Reducing and tolerating photon loss is the central technical challenge that will determine whether photonic approaches can work and which company will make it to commercial utility.

Performance Metrics — The Scorecard

The table below represents a snapshot of where photonic quantum computing stands in early 2026. The field is moving rapidly, and today’s headline numbers will be obsolete within 12–18 months. Different companies optimize different corners of this table: some focus on loss reduction, some on manufacturing scale, some on modular networking. Numbers should be read as directional benchmarks, not as grounds for declaring a winner.

Industry Landscape

Photonic quantum computing is unusual among quantum hardware sub-categories: it lacks a dominant incumbent of the IBM or Google variety, which means the race is genuinely open. The field is populated by a mix of growth-stage startups (PsiQuantum, Xanadu, Photonic Inc.), early-stage companies (ORCA, Quandela, QuiX Quantum), and larger corporate players with photonic research programs.

What’s distinctive about this landscape is the range of technical bets being made simultaneously. Some companies (PsiQuantum) are going “all in” on fault-tolerant, million-qubit systems using CMOS manufacturing — essentially skipping the noisy intermediate-scale era entirely. Others (Quandela, QuiX Quantum, ORCA) are deploying near-term commercial systems today, building a user base and ecosystem while the hardware matures. Xanadu sits in the middle: it has deployed real hardware (Aurora) and real software (PennyLane, the most widely used quantum ML framework), while publishing peer-reviewed milestones on the path to fault tolerance. This diversity of timelines and bets makes the photonic landscape harder to read than, others but there are multiple paths to potential commercial relevance.

Key Companies

Below is a non-exhaustive overview of current photonic quantum computing players and some highlights on each:

Alpha Dog — Xanadu

If I have to crown an Alpha Dog in photonic quantum computing today, I pick Xanadu. This is not the obvious choice — PsiQuantum has raised nearly 8× more money — but Xanadu has done something more important: it has shown its work, repeatedly, in peer-reviewed journals.

Xanadu’s flagship achievement in this cycle is Aurora — a 12-qubit modular photonic quantum computer built from four interconnected server racks, 35 photonic chips, and 13 km of fiber optics, all running at room temperature. Published in Nature in January 2025, Aurora is the first demonstration of a scalable, networked, modular photonic quantum computer with real-time error correction. Xanadu’s CEO called it the solution to the scalability problem in photonics: the modular architecture could, in principle, be extended to thousands of racks and millions of qubits without a fundamental redesign. Then, in mid-2025, Xanadu published a second Nature paper demonstrating 12 logical GKP qubits with real-time error correction — a meaningful step toward fault tolerance. And then a third Nature paper demonstrating the first on-chip generation of GKP states, enabling deterministic, room-temperature operations. That is three peer-reviewed Nature papers in a single calendar year, each building on the last, which is an extraordinary pace of publication and validation for a company of this size.

Xanadu is the first pure-play photonic quantum computing company to go public, listing on both Nasdaq and the Toronto Stock Exchange in Q1 2026, raising approximately $302M in gross proceeds. DARPA has advanced Xanadu to Stage B of its Quantum Benchmarking Initiative, recognizing it as one of a select group with a credible path to utility-scale quantum computing. Canada and Ontario are in discussions to provide up to C$390M for manufacturing infrastructure under Project OPTIMISM. On the software side, PennyLane, Xanadu’s open-source quantum ML framework, has become the de facto standard tool for hybrid quantum-classical machine learning, giving Xanadu an ecosystem moat that no other photonic company can match today. The combination of hardware milestones, public-company accountability, government backing, and software ecosystem is what sets Xanadu apart.

SWOT: Xanadu

Why I Picked Xanadu as Alpha Dog

They publish. Three Nature papers in one year, building on each other, with externally verifiable results. In a field prone to vaporware, peer-reviewed demonstration is the strongest possible pre-revenue signal of credibility.

Aurora solved the scaling architecture problem. Modular, room-temperature, networked, and already operating at 13 km of fiber — this is what a quantum data center actually looks like. The architecture is right, even if the qubit count is still small.

GKP qubits are the right long-term answer. The on-chip demonstration of GKP states is a founding-class milestone for the field. Xanadu is now the clearest technical leader on the error-correction roadmap for photonics.

PennyLane gives them an ecosystem that their hardware alone never could. Developers who learn quantum ML on PennyLane will naturally reach for Xanadu hardware when it is good enough. That’s a durable advantage.

Breakout Contender — Photonic Inc.

My Breakout Contender is Photonic Inc. (given its similar name with the field covered in this post, I’ll refer to them as “Photonics Inc.” to clarify the company versus the sub-sector), a Vancouver-based company building what it calls an “Entanglement First” distributed quantum computing system using a genuinely novel qubit: the silicon T-centre, a stable atomic defect in silicon that has both spin-based quantum memory and a native photonic interface operating at telecom wavelengths.

Photonic Inc’s bet is different from either Xanadu or PsiQuantum. It’s not trying to build a pure photonic quantum computer in the traditional sense; it’s using silicon spin qubits (the T-centre) as the memory and compute elements, and photons as the connective tissue that links those qubits together — both on-chip and across fiber-optic networks. This means Photonic Inc. gets the best of both worlds: the stability and fidelity advantages of spin qubits, and the networking and room-temperature advantages of photonics. Their architecture is designed from day one for distributed computing — linking modules across telecom-grade fiber rather than trying to cram everything onto a single chip or into a single cryostat. In January 2026, Photonic Inc. announced a $180M CAD ($130M USD) first close of a new funding round to accelerate toward product milestones.

The Microsoft partnership is the most significant signal here. In May 2024, Photonic Inc. demonstrated quantum entanglement transfer between two physically separated qubits using photons at telecom wavelengths — a milestone achieved in just six months of collaboration. Microsoft’s investment of time, engineering resources, and credibility in Photonic Inc. is a strong external validation of the architecture.

Differentiators

• T-centre silicon spin-photon qubit: Each T-centre has four spin qubits and an inherent optical interface at telecom wavelengths. This is not a workaround, it’s a purpose-designed qubit for distributed quantum computing. No other company has this specific qubit type.

• Native telecom wavelength networking: Operating at O-band telecom wavelengths means quantum information can travel through existing fiber-optic infrastructure without frequency conversion. For any application involving quantum networks, quantum-secure communication, or distributed computing, this is a huge structural advantage.

• DARPA QBI Stage B selection: DARPA has selected Photonic Inc. for its Quantum Benchmarking Initiative Stage B, evaluating it as one of a small number of platforms with a credible path to utility-scale quantum computing.

Why This Is One to Watch

The Microsoft partnership is underappreciated. If Microsoft, with its own deep quantum program (topological qubits), is investing engineering resources in Photonic Inc’s architecture, that says something. Watch for the collaboration to deepen, possibly including Azure cloud integration.

Distributed architecture is the correct long-term answer. Every other modality is trying to figure out how to network their qubits. Photonic Inc. designed for distributed computing from day one. If the field converges on distributed architectures as the path to millions of qubits, Photonic Inc. has a multi-year head start.

Milestones that would re-rate the company: A demonstrated multi-qubit logical operation across two networked modules; a public cloud integration (especially with Azure); or a peer-reviewed paper showing below-threshold error correction with T-centre qubits.

The $180M CAD raise in January 2026 gives them enough runway to hit several of those milestones. Watch for a larger round, possibly U.S. government participation, given their DARPA status, within 12–18 months.

If quantum networking becomes as important as quantum computing, which I expect, Photonic, Inc. is exceptionally well positioned to become a leading player in that field.

Where Photonic Quantum Computing Is Headed

On the technical side, three hurdles have to be met before photonic quantum computing moves from “compelling thesis” to “indispensable infrastructure.”

The first is optical loss reduction. Loss is to photonics what coherence time is to superconducting, namely the constraint that limits everything else. Xanadu has shown decent progress with their 60% loss reduction in 2025, but reaching the loss thresholds required for full fault tolerance will take more generations of chip fabrication improvements, better photon-number-resolving detectors, and optimized packaging. GKP qubits help by making the architecture more loss-tolerant, but the underlying photon loss must still come down significantly.

The second hurdle is demonstrating useful logical qubits at scale. Xanadu has 12 logical GKP qubits; PsiQuantum claims the manufacturing infrastructure to eventually reach millions of physical qubits. The gap between “demonstrated 12 logical qubits” and “running a commercially meaningful fault-tolerant algorithm” is immense — but it’s not qualitatively different from the gaps every other modality faces. The community is converging on the view that 1,000–10,000 high-quality logical qubits would unlock the first genuinely transformative applications in chemistry, materials science, and cryptanalysis. Getting there in photonics requires combining the manufacturing progress of PsiQuantum’s approach with the architectural validation of Xanadu’s approach.

The third hurdle is economic viability and business model clarity. Photonic systems are attractive because they could be cheap at scale, manufactured in CMOS fabs, deployed at room temperature, and networked over existing fiber. But “could be cheap at scale” and “is cheap today” is not the same thing. The near-term revenue models are dominated by government contracts, HPC integration (Quandela at the TGCC supercomputer), and early enterprise pilots. The inflection happens when recurring commercial workloads justify dedicated quantum hardware budgets , but there is likely a long gap (measured in years) before that happens.

The Elephant in the Room - PsiQuantum

No discussion of quantum computing is complete without addressing the largest single bet in the industry. PsiQuantum has raised approximately $2.37 billion. They have broken ground on datacenter-scale facilities in Chicago and Brisbane, partnered with GlobalFoundries, published the Omega chipset paper in Nature, and partnered with NVIDIA. These are real milestones. But PsiQuantum operates with a level of opacity that is extraordinary even by quantum industry standards. The company has not published performance data on a working multi-qubit system, has not provided public benchmarks on a deployed processor, and has not announced any external access program.

This matters for the whole sector. If PsiQuantum’s manufacturing-first, fault-tolerant thesis is validated, if they can demonstrate that CMOS-fab-produced photonic chips can be assembled into a functioning large-scale quantum computer, it would be one of the most significant technology events of the decade, and would likely validate the entire photonic modality with a force multiplier effect on every other company in the space. But if PsiQuantum’s approach hits fundamental obstacles that have been hidden behind the veil of secrecy, or if the timelines slip years beyond current targets, the fallout would extend well beyond just PsiQuantum. At $2.37B raised and a $7B valuation, their failure would be the largest single capital destruction event in quantum hardware history, and it would almost certainly suppress valuations and funding appetites for photonic (and possibly all) quantum hardware for years. That is a real systemic risk for the quantum market that every investor and ecosystem participant should watch carefully.

Signals I’m Watching for Photonic Quantum Computing Over the Next 1–3 Years

• PsiQuantum publishes a multi-qubit system benchmark. Any credible, externally verifiable performance result from a working PsiQuantum system, even a small-scale one, would be a major signal. The longer the silence, the bigger the question mark. Conversely, even a modest public demonstration would be enormously de-risking for the entire sector.

• Xanadu’s loss reduction trajectory. If Xanadu’s optical loss continues to fall at the rate seen in 2025, they reach the thresholds for fault-tolerant operation within 2–3 years. If progress slows, the timeline extends significantly. Watch for their quarterly/annual technical reports and loss metric updates.

• First integration of a photonic quantum processor into a commercial HPC workflow. Quandela’s deployment at the TGCC supercomputer is a prototype of what this might look like. The first case where a photonic QPU is genuinely part of a production computational workflow, with SLAs, recurring usage, and measurable value, is a milestone that changes buyer behavior across the market.

• Photonic Inc. + Microsoft roadmap announcements. If Microsoft announces Azure quantum integration with Photonic Inc’s architecture, or if Photonic Inc. publishes results on multi-module entanglement, the “distributed quantum computing” narrative gets a large institutional validation.

• Competitor modalities closing the networking gap. If neutral atoms or superconducting systems demonstrate scalable, fiber-compatible quantum networking, photonics loses one of its most distinctive advantages. Watch for Google’s dual-modality strategy and neutral-atom networking experiments specifically.

From a full-stack perspective, photonic quantum computing is uniquely positioned to serve as both the computing layer and the networking layer of the future quantum stack. Every quantum modality will eventually need photons to communicate, whether that’s linking quantum computers across a campus, distributing entanglement for quantum cryptography, or connecting quantum sensors. The companies that crack photonic computing are likely also building the hardware for the quantum Internet. That is a much larger addressable market than quantum computing alone, and it’s part of why the capital concentration in photonics, despite the formidable technical challenges, makes strategic sense.

Disclosure: The author is a venture investor with investment interests in quantum and may have an interest in companies discussed in this post. The views expressed herein are solely the views of the author and are not necessarily the views of Corporate Fuel Partners or any of its affiliates or any companies it has investment interests in. Views are not intended to provide and should not be relied upon for investment advice.

References:

“Five Reasons Photonic’s Quantum Architecture Stands Apart,” Amadeus Capital, January 7, 2026.

“How Much Did PsiQuantum Raise? Funding & Key Investors,” Clay, April 20, 2026.

Lenahan, B., “PsiQuantum’s FTQC Chase,” Quantum’s Business (Substack), December 16, 2025.

Lubiani, A. et al., “Scaling and networking a modular photonic quantum computer,” Nature, February 4, 2025.

Microsoft Azure Blog, “In Collaboration with Microsoft, Photonic Demonstrates Quantum Entanglement,” Microsoft, May 29, 2024.

“PsiQuantum Announces Omega, a Manufacturable Chipset for Photonic Quantum Computing,” PsiQuantum Press Release, February 25, 2025.

“PsiQuantum Breaks Ground on America’s Largest Quantum Facility,” PsiQuantum Press Release, September 29, 2025.

“PsiQuantum Raises $1 Billion to Build Million-Qubit Scale, Fault-Tolerant Quantum Computer,” PsiQuantum Press Release, September 9, 2025.

“PsiQuantum Valued at $7 Billion in Latest Funding Round, Teams Up with Nvidia,” Reuters, September 10, 2025.

“Photonic Quantum Computers Market Report 2025: Room-Temperature Quantum Computing Gains Traction,” BusinessWire, July 28, 2025.

“Photonic Quantum Computing Market Research Report 2034,” Researchintelo, April 8, 2026.

“Photonic Raises $180M CAD to Accelerate Distributed Quantum,” Photonic Press Release, January 5, 2026.

“Quandela to Launch Belenos, Accelerating its 2030 Roadmap toward Scalable Photonic Quantum Computing,” The Quantum Insider, May 21, 2025.

Quandela / CEA, “Quandela Delivers Photonic Quantum Computer to EuroHPC and GENCI at CEA’s TGCC,” The Quantum Insider, October 22, 2025.

“QuiX Quantum Secures €15 Million to Deliver First-Generation Universal Photonic Quantum Computer,” QuiX Quantum, July 10, 2025. “Xanadu Announces Fourth Quarter and Full Year 2025 Results,” GlobeNewswire, April 9, 2026.

“The Global Photonic Quantum Computing Market 2026–2036,” Future Markets Inc., March 22, 2026.

Vernon, Z. et al., “First on-chip photonic qubit enables GKP states for error correction at room temperature,” Nature, June 2025, covered at phys.org.

“Xanadu Becomes First Pure-Play Photonic Quantum Computing Company to Go Public,” PRNewswire, March 26, 2026.

“Xanadu Introduces Aurora: World’s First Scalable, Networked and Modular Quantum Computer,” PRNewswire, January 21, 2025.