DARPA just cut its quantum computing contest down to four teams.
The Defense Advanced Research Projects Agency announced Wednesday that Atom Computing, Diraq, IBM and IonQ have advanced to the final stage of its Quantum Benchmarking Initiative. The program exists to answer one blunt question: can anyone build a quantum computer that's actually useful by 2033, seven years from now. Each of the four finalists is betting on a different kind of qubit, the basic unit of quantum information, so the agency is effectively running four competing bets on which physical approach wins. DARPA hasn't said how many teams entered earlier rounds or what happens to the ones that didn't make this cut.
Quantum computing has been "five to ten years away" for most of the last two decades, and DARPA's benchmarking program is one of the few efforts trying to pin that estimate down with an actual deadline and a government customer attached. That four different qubit architectures are still in contention this late in the process is itself notable - it suggests the industry hasn't converged on a winning hardware approach the way classical computing did around silicon transistors.
A 2033 deadline sounds comfortably far off until you remember IBM has been promising practical quantum advantage on a rolling basis since at least 2019.