This week the news worth reading wasn't a qubit count — it was the plumbing. Three announcements, all about the unglamorous engineering that fault tolerance actually runs on. Quobly, the Grenoble silicon-spin startup spun out of CEA-Leti and CNRS, closed a €115M ($133.5M) Series A to build its first machine, Alloy Pioneer, on the same FD-SOI 300mm wafers that ordinary chips already use. There's no qubit count yet and no published gate fidelities, so this is a manufacturing bet, not a hardware result — but it's the right bet: if silicon-spin qubits work at all, the whole reason to care is that you could print them in an existing CMOS line instead of hand-building them one at a time. QuiX Quantum, in the Netherlands, installed something even less photogenic and more important — a feed-forward control unit that reads a single-photon detector and reconfigures the optical chip in about 150 nanoseconds, before the photon has traveled 30 meters down the fiber. Measurement-based photonic quantum computing simply doesn't work without real-time feedback this fast, so this is a real piece of the fault-tolerance puzzle clicking in, not a press-release number. And Hamamatsu Photonics, NKT Photonics, and Yaqumo signed a Japan–Denmark agreement to standardize the lasers, detectors, and optical modules that cold-atom machines need — the quiet supply-chain work that decides whether any of this scales past one lab bench. None of the three is a breakthrough in the physics. They're all engineering, and engineering is where the entire game sits right now. We'll take three boring announcements about wafers, FPGAs, and lasers over one more chart of rising qubit counts.
3件のリンク