Initiating Outperform: Pole Position in Quantum Race

研报英文原文证据摘录

Initiating Outperform: Pole Position in Quantum Race

erstate its claims?

There is no standard definition industry-wide for what characteristics a qubit has to have in order

to be “logical” instead of just “physical”. To-date, its meaning depends on the error-correcting

EVR ISI View: Under the code, the code distance, and the physical to logical overhead – and the bar ranges from

strictest full-fault tolerance “demonstrated below-threshold error suppression” at the strict end to just a number on a 2030

(below-threshold) definition that roadmap at the loose end. Our sense is the purist definition would be that a logical qubit is full

Google demonstrates and IBM fault-tolerant with error-rates that would continue to fall as physical qubits are added. QNT’s “fully

targets, Helios’s 48 logical qubits error-corrected” logical qubits are encoded with high-rate “iceberg” codes at a 2 to 1 rate,

– built on high-rate, distance-2- catching a single error rather than correcting an arbitrary one. However Helios’s logical qubits

class “iceberg” definition codes operate “beyond break-even” with error rates 10x to 100x below their physical counterparts.

at a 2:1 ratio – would not all Moreover, as of June 2025, QNT has separately made the claim to have demonstrated a fully

qualify as “true” fault-tolerant fault-tolerant universal gate set with repeatable error correction – so one could argue it can do

logical qubits. But QNT logical the strict version, it just uses an efficient code to maximize the Helios count.

error rates are 10x to 100x lower

vs physical and has separately IONQ’s #AQ algorithmic qubit metric is not a logical qubit count at all. The metric is inspired by

demonstrated fully fault-tolerant the QED-C Technical Advisory Committee’s application-oriented benchmark.