Inside D-Wave’s Dual-Rail Qubit Roadmap and the New Haven Bet on Logical Qubits

D-Wave spent two decades as the company that did not build gate-model machines. It made its name on quantum annealing, sold annealing systems commercially, and largely sat out the race that IBM, Google, Quantinuum and a string of startups were running toward universal, gate-based quantum computers. That posture has now reversed, and it reversed through an acquisition rather than a pivot in D-Wave’s own labs.

The engine of the new gate-model programme is a research centre in New Haven, Connecticut, and almost everything D-Wave presented on gate-model hardware in 2026 traces back to it. That trail runs from the roadmap it published at its June Investor Day to the Nature paper that landed in August. The bet those slides describe is not a bigger qubit count but a cheaper one. The first hardware that can test it is DR-17, which D-Wave’s June roadmap dates to 2026, and its results release of 6 August repeated that date.

D-Wave’s gate-model bet is built on the dual-rail qubit. The architecture arrived through its purchase of Quantum Circuits Inc, announced at $550 million and completed on 20 January 2026, and with it the New Haven team of transmon co-inventor Rob Schoelkopf. It was bought, not grown in D-Wave’s own annealing labs.