PsiQuantum builds photonic quantum sites in Chicago and Australia

For most of the last decade, quantum computing has been the technology that is always five years away. A profile in MIT Technology Review of PsiQuantum, the photonic-qubit company that raised $1 billion last year and is slated to receive $100 million in CHIPS Act funding, is the clearest signal yet that DARPA and a growing chunk of the industry now believe the timeline is real.

The company is betting on light. Google and IBM are pursuing superconducting qubits and Intel is using electrons, but PsiQuantum routes photons through chips of its own design. It has broken ground at the Illinois Quantum and Microelectronics Park outside Chicago and on a second site in Moreton Bay, Australia, with chips made by GlobalFoundries in Malta, New York. The Australia facility is promised operational, meaning hardware-ready, in 2027, which is not the same thing as a working million-qubit machine switched on that day.

Why DARPA matters here more than the funding: the Pentagon's evaluation has been scrutinising PsiQuantum's systems since 2023, and last year placed the company into the third stage of a benchmarking initiative that only PsiQuantum and Microsoft have reached. Joe Altepeter, who ran the program, said he is 'more optimistic now than I have been at any point in the past 10 years', and his successor Micah Stoutimore now says it 'seems likely that someone will build a utility-scale quantum computer by 2033'. Coming from the agency paying to stress-test the field, that is not marketing copy.

The honest caveat is that, as UT Austin's Scott Aaronson tells the magazine, 'it is very hard for an outsider to evaluate'. Independent researchers Andrew Childs and Dominic Berry, quoted in the piece, warn that the speedups shown so far are unlikely to have significant practical impact until systems are much larger, and that impressive chemistry improvements would still need faster algorithms to answer the questions scientists most want to answer. What the reporting does not give you is the current error rate inside the three connected cabinets at Milpitas, or the gap between 'hardware-ready' in 2027 and a real machine doing real work for real customers.

If PsiQuantum is right, the winners are the partners already inside the tent, Lockheed Martin on materials, Mercedes on batteries, and Airbus on aerospace applications, plus GlobalFoundries picking up a second architecture to fabricate at scale. If it is wrong, the DARPA shortlist itself is still the datapoint worth watching, because the field's referee just told everyone how many horses are actually left in the race.