PsiQuantum's Million-Qubit Fab Strategy

PsiQuantum is trying to win quantum computing the way TSMC won chipmaking, by turning a fragile lab device into a repeatable factory process. The bet is that useful quantum machines will need error correction and at least a million qubits, so the hard problem is not adding a few more qubits to a demo chip, it is making photon sources, detectors, and optical switches in volume on standard 300mm wafers with a foundry partner, then stitching those parts into a system that can actually be manufactured repeatedly.

• This makes PsiQuantum look less like a science project and more like a semiconductor program. The company says it now manufactures millions of waveguide integrated photon number resolving detectors at GlobalFoundries, and has installed dedicated production equipment in GF fabs to produce thousands of photonic chips, which is a very different milestone from a record qubit count on a single prototype.
  2 psiquantum 3 psiquantum
• The Xanadu comparison shows why this matters. Both use silicon photonics and foundry partners, but Xanadu is built around continuous variable squeezed light systems while PsiQuantum uses single photon qubits. PsiQuantum is putting more of its strategy weight on standard fab flows, telecom fiber links, and manufacturable components, which fits a direct path to fault tolerant scale if the hardware modules can be yielded reliably.
  4 psiquantum
• Capital has followed the manufacturing thesis. By September 2025, PsiQuantum had raised a $1 billion round at a $7 billion valuation, and recent reporting ties that financing to building utility scale sites in Chicago and Australia. That funding is less about near term access revenue and more about paying for foundry process development, packaging, facilities, and long lead infrastructure.
  1 sacra 6 psiquantum 7 sacra

The next phase is where quantum starts to resemble advanced chip production. If PsiQuantum keeps converting custom photonic components into standard wafer and packaging steps, the company can compound manufacturing learning faster than rivals that are still centered on small system demos. That would shift the market from a race for headline qubit counts to a race for yield, supply chain depth, and deployable fault tolerant systems.