Google Quantum AI publiera-t-il un article revu par les pairs revendiquant le franchissement du Milestone 3 avant le 31 décembre 2027 ?

• Ce qui fait monter. Google's official blog announced a major breakthrough in a quantum computing 'bottleneck' on 2026-10-15 [1].
• Ce qui fait monter. Reports indicate Google is already demonstrating 'Magic State Cultivation' on the Willow processor, a step that typically follows or coincides with achieving Milestone 3 [3].
• Ce qui fait monter. The Willow processor (105 qubits) has already demonstrated below-threshold error correction, providing the necessary hardware foundation for M3 [1, 5].
• Ce qui fait monter. Google's 2029 target for a useful quantum computer necessitates reaching M3 by 2027 to allow time for scaling and logical operations [1].
• Ce qui fait descendre. The peer-review process for major scientific papers can take 6 to 12 months, meaning the experiment must be completed by early 2027 at the latest [2].
• Ce qui fait descendre. Milestone 3 requires the logical qubit to perform 'substantially better' than the best physical qubit, a high bar that has historically proven difficult to clear [1].
• Ce qui fait descendre. The transition from M1 to M2 took nearly five years, suggesting that progress in error correction is slower than hardware iterations [1].
• Ce qui ferait le plus bouger ce chiffre. Whether the 'bottleneck' breakthrough announced in October 2026 refers specifically to the break-even point required for Milestone 3 or a different technical optimization.

• Ce qui fait monter. Google's demonstration of below-threshold error suppression on the Willow chip (Milestone 2) established the necessary physical error rates to make a net-positive logical qubit feasible [tier 1].
• Ce qui fait descendre. With only 14 months remaining until 31 December 2027, the standard 6 to 12-month peer-review publication pipeline at journals like Nature or Science leaves a very narrow window if a paper has not already been submitted [tier 1].
• Ce qui fait descendre. The step between Milestone 1 (2019) and Milestone 2 (2023–2024) took nearly five years, making peer-reviewed delivery of Milestone 3 by late 2027 significantly faster than Google's prior cadence [tier 1].
• Ce qui ferait le plus bouger ce chiffre. Whether Google has already submitted a manuscript demonstrating break-even logical memory or will need another hardware spin beyond Willow to achieve substantial lifetime improvement.

• Ce qui fait monter. Google’s recent expansion into neutral‑atom qubits suggests potential acceleration of error‑correction progress [tier 4]
• Ce qui fait descendre. No peer‑reviewed milestone 3 result yet; Google’s own timeline targets 2029 for a useful machine, implying a 5‑year gap similar to M1→M2 [tier 1]
• Ce qui fait descendre. The lack of any announced or published milestone 3 evidence in the last year pushes the probability down [tier 4]
• Ce qui ferait le plus bouger ce chiffre. Whether Google’s neutral‑atom program will deliver a logical qubit with lifetime exceeding the best physical qubit before the end of 2027.

• Ce qui fait monter. Google has successfully validated M1 and M2, demonstrating a track record of hitting quantum milestones, though the time gap between M1 and M2 was ~5 years [tier 1 context].
• Ce qui fait monter. The opening of a second modality (neutral atoms) in March 2026 suggests accelerated R&D capacity [tier 1 context].
• Ce qui fait descendre. The hand-verified state explicitly states M3 remains an active research target with no peer-reviewed publication as of the current date [tier 1 context].
• Ce qui fait descendre. Google's official target for a 'useful machine' is 2029, suggesting M3 (a key step toward utility) is unlikely to be fully published and peer-reviewed by end of 2027 [tier 1 context].
• Ce qui fait descendre. The provided recent press articles are mostly tier 4 aggregators or unrelated tier 2 news, providing no new tier 1 or 2 evidence of imminent M3 breakthroughs [tier 2/4].
• Ce qui ferait le plus bouger ce chiffre. Whether the 'substantially better' lifetime threshold for M3 has been achieved in internal labs but is being held back for a larger 2028/2029 announcement, or if the technical barrier is still unsolved.

Raisons non conservées pour cette révision (l'historique détaillé n'est gardé que depuis peu).