Careful though-- physical qubits ≠ progress. Actual progress is 🔥 The QPU industry's actual failure isn't the trajectory it's the access model. Research can't get QPU time without piles of money. (That's what @QuantumSN48 fixes on @bittensor). For the geeks-- a few important notes: → Dec 2023, IBM Condor hit 1,121 qubits. Missing from your chart. Same summit, IBM also announced Heron at 156 qubits with way better fidelity. Condor was a feasibility milestone, not a production target. Translation: industry decided quality is the new game. The whole industry pivoted that quarter. Charting physical qubits after that is charting a metric the field deliberately abandoned. IBM → Dec 2024, Google Willow crossed below-threshold error correction. First time in the field's 30-year history. A 101-qubit distance-7 code with logical error suppressed by Λ = 2.14 per code-distance increase, exceeding the best physical qubit's lifetime by 2.4x. Translation: Adding qubits now makes errors better, not worse. This is a key inflection. → Microsoft + Quantinuum: 12 logical qubits running with circuit error rate of 0.0011 vs. 0.024 for physical — 22x improvement. Translation: Logical isn't theoretical anymore. → Oxford Ionics: 99.99% two-qubit gate fidelity in 2025. Quantinuum: 99.9%+. Why is this important: Its squarely in the regime where error correction actually works. → RSA factoring requirement: dropped from 20 million qubits to ~1 million in 2025. Goalpost moved 20x closer in a single paper. → Quantinuum Helios shipped commercially Nov 2025 — 98 fully-connected trapped-ion qubits, >99.9% fidelity, with a new real-time control engine. This is production hardware, today. Right metrics to watch are logical qubits and gate fidelity. Both are on a tear. Charting physical qubit count past 2023 is like charting CPU GHz past 2005 (measuring the metric the industry deliberately stopped optimizing)