English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Who Gets to 100 Logical Qubits First: A Roadmap Showdown (Nov 2026 Perspective)

Forum topic · 小凯 · 2026-11-07

Summary

This article compares the five major qubit technology platforms—superconducting (IBM), trapped ions (Quantinuum), neutral atoms (Harvard-based teams), topological (Microsoft), and silicon spin (Intel)—by their pace toward the industry-accepted milestone of 100 logical qubits with error rates around 10⁻⁶. It explains why 100 logical qubits marks the boundary between demonstrations and serious computation, noting the physical-qubit overhead of ~200,000 for surface codes, halved by qLDPC codes. Key reading rules are offered: distinguish physical from logical qubits, watch trajectory slope rather than snapshots, and scrutinize the full engineering stack beyond chip announcements. As of the November 2026 perspective, trapped ions and neutral atoms lead on quality records and steep improvement curves, superconducting platforms chase via modularity and new codes but face calibration bottlenecks, topological qubits remain a wildcard, and silicon spin is positioned for the million-qubit era.

This article is part of a Chinese tech forum's quantum computing column ("From Coins to Error Correction"), installment #070, closing its "Hardware Deep Dive" series. It frames the race among five qubit platforms as a marathon pacing chart, asking: who reaches 100 logical qubits first?

Why 100 Logical Qubits Is the Benchmark

The milestone is a consensus anchor, not an official rule. With surface codes, one logical qubit costs roughly 2,000 physical qubits—so 100 logical qubits implies about 200,000 physical qubits. Switching to newer qLDPC codes can halve that bill. "100 high-quality logical qubits" means error-correction overhead becomes tolerable and Shor-scale deep circuits become feasible—the dividing line between "running serious problems" and "demonstrations." The judging criteria have two variables: quantity (100) and quality (logical error rates on the order of 10⁻⁶).

The Five Platforms' Pacing Chart

  • Superconducting (IBM) — the large army: Largest physical qubit scale (thousand-qubit chips released), two-qubit fidelity hovering around 99.9%. Strategy: modular architecture + code switches to reduce overhead. Biggest risk: it alone faces the "three walls" of cryogenic capacity, wiring, and calibration. Pace: "most people, most guns, heaviest logistics."
  • Trapped ions (Quantinuum) — elite squad: Highest gate fidelities (above 99.9%), frequent holder of logical-qubit quality records. Weaknesses: chains can't be too long, slow gates; scaling relies on modular approaches. Pace: "few but accurate, slowest marching."
  • Neutral atoms (Harvard-affiliated team) — the promoted newcomer: Current holder of the 48-logical-qubit record; reconfigurable atom arrays give code-design freedom. Weaknesses: atom loss, still catching up in fidelity. Pace: "steepest slope—from rookie to leader in four years."
  • Topological (Microsoft) — the high-stakes gambler: Evidence for Majorana modes remains contested; shallowest scale demonstrations. If confirmed, "patrol-free" qubits skip the error-correction arms race. Pace: "no pacing chart visible (track not yet certified)."
  • Silicon spin (Intel) — the late-stage runner: Highest density ceiling, existing fabs, but smallest-scale demonstrations. Pace: "conserving strength early, betting on factory muscle in the back half."
  • Three Rules for Reading Quantum Headlines

    1. Physical qubits ≠ logical qubits: judge by logical qubits and logical error rates, not raw counts. 2. Watch slope, not snapshots: today's leader isn't necessarily first across the line—neutral atoms' four-year trajectory vs. trapped ions' fifteen-year accumulated quality advantage both matter. 3. Look at the internal organs: half a machine is control systems and wiring. Discount announcements that tout chips but omit the engineering stack by about 30%.

    The Betting Sheet (Nov 2026 View)

  • First tier: trapped ions and neutral atoms—quality records in hand, steep slopes; likely to touch the 100-logical-qubit threshold within two to three years.
  • Superconducting: could catch up in the same window via modularity + new codes, with the calibration wall as the biggest wildcard.
  • Topological: the black swan—if evidence solidifies, the whole timetable is rewritten.
  • Silicon spin: not a sprinter this round, but a leading candidate for the next round (million-qubit era), offering another answer to the manufacturing wall.
  • Where the Marathon Analogy Breaks Down

  • In a marathon, everyone runs the same track under the same physics; the five quantum platforms are different physical systems whose "weights" for speed, fidelity, and scale differ—there's no accepted conversion formula for who is truly "faster."
  • The marathon finish line is fixed; "100 logical qubits" is a consensus anchor, not a legal finish, and a 100-logical-qubit machine at 10⁻⁴ error rates is worth less than 50 logical qubits at 10⁻⁶.

Takeaways

1. Dual metrics: 100 logical qubits + 10⁻⁶ error rate is the threshold band for "serious computation." Convert physical-qubit headlines before believing them. 2. Pacing-chart literacy: slope over snapshot, logical over raw, engineering stack over chip close-up—the readings don't expire, the paces do. 3. The five-way landscape (Nov 2026): quality duo (ions/neutral atoms) leading, superconducting army chasing, topological tiger crouching, silicon spin betting on the next round.

The post closes the "Hardware Deep Dive" batch and previews the next series on quantum information ecosystems—who uses these machines, networking, cryptography, and standardization. A related link on the forum covers Quantinuum partnering with Quanta Computer for quantum manufacturing.

Tags

#quantum-computing#logical-qubits#error-correction#superconducting-qubits#trapped-ions#neutral-atoms#topological-qubits#technology-roadmap

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178636904