Could you start by introducing Photonic and explaining what the company does?
Photonic Inc. is at the forefront of quantum technology, building commercial-scale quantum computers and networks to tackle big challenges like materials science, drug discovery, climate change, and security.
What sets Photonic apart is its ability to scale performance through advanced distributed quantum computing. At the heart of this is the Entanglement First™ architecture, which uses optically linked silicon spin qubits for exceptional connectivity. This approach enables powerful computation, efficient error correction, and seamless integration with today’s data centers and telecom systems.
Photonic is a 160-person company headquartered in Vancouver, British Columbia, with subsidiaries in the US and the UK, and about 40 percent of our people working remotely around the world.
Let’s unpack that a bit. Entanglement and teleportation sound like science fiction. Can you explain what they really are and how they change computing and networks?
The transistor already used one quantum property – electron tunnelling – so classical computing is already sitting on quantum foundations.
What we haven’t fully used until now are the other two quantum properties: superposition and entanglement. If you only use transistors, you’re a regular computer company. If you add superposition, you’re getting into being a quantum company. Once you master entanglement, you’re a fully quantum company – because entanglement is the currency of quantum computing.
Classical computing has a few basic gates; quantum computing has many more, and you make those gates by creating and consuming entanglement. So the core question for any quantum company is: how do you make it, and how do you consume it?
People we speak to often mention telecom-connected silicon spin qubits as Photonic’s key differentiator. When you imagine this model at scale, why is it better – and why can it truly scale versus trapped ions or superconducting approaches?
I came out of building very large systems – 20 years in wide-area networks, helping build what became the internet, and 10 years in supercomputing at Cray Canada. So I care a lot about scale, and to me scale means the marginal cost of adding one more unit should tend to zero. If it doesn’t, your technology will get big, but it won’t really scale. The internet scaled because adding a router cost almost nothing and suddenly the whole thing was bigger. That’s the mindset.
Back in 2016 I was actually a quantum skeptic because I saw three hard problems no one had solved. Then I met our founder, Stephanie Simmons, and she was dedicated to solving them. What she found was effectively a “transistor for quantum”: a defect in silicon – a T-centre – that can hold four qubits, can be controlled for superposition, can store quantum states (giving you memory), and, crucially, emits light at telecom/data-centre wavelengths. That means you can link two of these objects over existing fibre and entangle qubits that have never sat next to each other. You avoid proximity-based challenges, and the marginal cost of adding qubits drops toward zero – that’s why it scales.
You mentioned that your way of doing entanglement is your unique differentiator. Can you walk us through that and why it matters for building a fault-tolerant machine?
Most quantum companies do proximity-based architectures: put qubits very close together so they can entangle. The problem is, the closer they are, the more quantum crosstalk you get, which is bad.. What we do instead is take two T-centres that don’t have to be on the same chip – they can theoretically even be in different countries – make each of them emit a single photon, interfere those photons on a beam splitter, and project the entanglement back onto the spins that emitted them. You’ve now entangled two qubits that have never met. That idea won the Nobel Prize in 2022, and we’re the only commercial outfit doing it this way.
Because we can print large numbers of these T-centres on silicon, connent them, and have every pair entangle on demand, we get a highly parallel, GPU-like quantum architecture. And because our qubits can all be connected, we can use quantum LDPC error-correction codes – which we announced this year – instead of needing 10,000 physical qubits per logical qubit like some surface codes do. In our networked architecture you can get to 10–30 physical qubits per logical qubit. That’s the difference between a science project and something that can reach commercial scale.
So if we shift the tone to the curious investor: what does the roadmap look like over the next few years, and what does success look like by 2030?
I think the right way to think about the market is: by 2030 you want roughly 100,000 logical qubits in service, however the industry achieves that. Most real applications people care about – chemistry, materials, catalysts, biotech, some AI training-data generation, and later finance and combinatorial optimisation – sit in the 100 to 2,000 logical-qubit range. They don’t need a quantum computer forever; they need it for a second, a minute, an hour. So you expose that capacity as a cloud service – that’s what our relationship with Microsoft is about.
Once you can deliver 100–2,000 logical qubits reliably, whole industries move: chemistry can collapse 1,000 years of discovery into a year; biotech can simulate reactions directly instead of approximating them; and at around 1,800 logical qubits you start to hit meaningful finance problems. That’s what success looks like: real logical qubits, delivered as a service, doing real workloads.
