PsiQuantum

Overview

PsiQuantum is a private quantum computing company pursuing a singular, high-conviction bet: that fault-tolerant, utility-scale quantum computing can only be achieved at the physical qubit counts required by building on standard semiconductor manufacturing infrastructure. Founded in 2016 by Jeremy O'Brien, Terry Rudolph, Mark Thompson, and Pete Shadbolt — all photonics researchers from the University of Bristol — the company has deliberately avoided shipping near-term noisy intermediate-scale quantum (NISQ) hardware, instead targeting a first-generation machine requiring approximately one million physical qubits to support a meaningful number of error-corrected logical qubits. This approach places PsiQuantum in a distinct strategic category: it is not competing for today's cloud quantum revenue but for the first generation of machines capable of solving classically intractable problems of commercial significance.

The company's core technical thesis rests on photonic qubits — encoded in single photons — manufactured using silicon photonics processes compatible with existing CMOS semiconductor fabs. The central argument is that no other qubit modality can plausibly reach the physical qubit counts required for fault-tolerant computing without a manufacturing pathway that leverages the scale, yield improvements, and capital base already invested in the global semiconductor industry. PsiQuantum has formalized this thesis through deep partnerships with GlobalFoundries, one of the world's largest contract semiconductor manufacturers, and SkyWater Technology, a U.S.-based foundry with ties to government-backed domestic chip manufacturing initiatives. These partnerships are not peripheral: they are the load-bearing element of PsiQuantum's commercialization strategy.

Commercially, PsiQuantum is not generating product revenue and does not expect to do so until a fault-tolerant system is operational. Its near-term commercial relationships are structured around government contracts and strategic partnerships rather than hardware sales. Notably, the company secured substantial commitments from the Australian federal and Queensland state governments in 2023, totaling approximately AUD 940 million (roughly USD 620 million at the time), to establish a quantum computing facility in Brisbane — one of the largest government quantum investments globally. This positions PsiQuantum as a nationally strategic asset in Australia's quantum ambitions, reducing near-term funding risk while adding geopolitical and execution complexity.

In the competitive landscape, PsiQuantum occupies an unusual position. It is neither a NISQ-era commercial competitor to IBM, Google, or IonQ, nor a purely academic research program. Its closest philosophical analog is perhaps Google's long-horizon fault-tolerance roadmap, but PsiQuantum is entirely private, lacks a revenue-generating cloud business to subsidize R&D, and has staked everything on a single technical architecture. The company competes indirectly with photonic peers Xanadu (continuous-variable photonics, very different architecture) and QuiX Quantum, and more broadly against any company that achieves fault-tolerant quantum advantage before PsiQuantum ships hardware. The risk of being outpaced by superconducting or trapped-ion approaches achieving fault tolerance first is the defining existential question for the investment thesis.

Leadership

Technology

PsiQuantum's architecture is based on linear optical quantum computing using single photons as qubits, implemented in silicon photonic integrated circuits manufactured at commercial semiconductor foundries. The company's specific approach — called Fusion-Based Quantum Computing (FBQC), developed principally by co-founder Terry Rudolph and collaborators — differs from earlier linear optical schemes (such as the KLM protocol) by using probabilistic entangling operations called 'fusions' between small clusters of photons (resource states) rather than attempting deterministic two-qubit gates. This architecture is designed to be inherently tolerant of photon loss and gate failures, which are the dominant error modes in photonic systems, by spreading logical information across many physical photons and relying on large-scale resource state generation and measurement.