Finland – October 24, 2026 -- SemiQon, a 2023 spinout from Finland's VTT research institute, has shipped its first 4-qubit silicon spin-qubit chips to research collaborators after completing several fab runs. CEO and co-founder Himadri Majumdar says buyers should judge quantum processors on fidelity at scale and other system-level metrics, not on small-device fidelity alone.
Majumdar says 98% fidelity can be offset by qubit volume
Majumdar argues that some players over-optimize fidelity on small qubit counts. In his view, a sufficiently large number of physical qubits makes six nines of fidelity unnecessary. He puts the industry's physical-to-logical qubit ratio at about 100 to 1, down from 1,000 to 1, while some players promise 12 to 1.
Five metrics beyond fidelity shape scalability
Majumdar lists computation speed, multiqubit networking, control of individual qubits at scale, cooling and environmental control, and manufacturing. He notes that IBM's quantum volume metric does not describe computational power across multiple chips well. That gap matters as manufacturers reach the limit of superconducting qubits per chip in a single dilution refrigerator.
He adds that end users care only about solving industry problems in a commercially viable way. SemiQon calls silicon the only platform that can scale to billions of qubits and put classical control electronics on the same chip as the qubits.
Three-step wafer testing relies on a statistical subset
Each fab wafer yields thousands of chips. SemiQon first probes them at room temperature to remove malfunctioning chips. Cryogenic wafer-level characterization is the ideal next step, Majumdar says, citing a recent Intel paper and independent work at the University of Basel, a SemiQon collaboration partner.
The company then dices the wafer and benchmarks a random set of pre-screened chips on charge noise, pinch-offs and transistor behavior, extrapolating to the rest.
Chips reach four universities as a shared benchmark
Samples have gone to the University of Basel, Seoul National University, Taiwan National University and the University of Jyväskylä. Labs repeat the measurements to confirm results hold despite differing equipment. SemiQon plans to ship more samples and says wide distribution will create a benchmark for its chips' performance.
The work was funded under the EU's Horizon EIC programme, grant agreement No. 101136793 (SCALLOP). The source article was a sponsored post.