{"id":"f1cc99f9-146e-4b87-81cf-3d97f5b53c9f","arxiv_id":"2608.05010","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A systematic review of neutral-atom quantum computing covering principles, technology routes, progress from 2000 to 2026, industrialization, and core bottlenecks.","lead":"This paper reviews the principles, technical routes, and 2000 to 2026 progress of neutral-atom quantum computing, including Rydberg gates, atom arrays, and error correction demonstrations. It is a reference survey, useful for researchers and funders mapping the field's status, bottlenecks, and industrial landscape.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The review's quantitative map is built on two incompatible two-qubit fidelity baselines: §4.3/Table 1 report 99.84% as the max, while §5.1 and §6.1 treat 99.5% as the current ceiling for bottleneck analysis.","rationale":"The reader's weakest-assumption query was about the accuracy of secondary sources; this stress-test agrees partially but locates a more precise failure mode: the review cites the same body of literature inconsistently. The 99.84% versus 99.5% discrepancy is internal, concrete, and directly controls the main technical conclusion—how far current fidelity sits from the ~99.9% level needed for low-overhead fault tolerance. It can be settled by re-reading two published papers and re-deriving one table row, without new experiments. This concern reinforces rather than overturns the reader's CONDITIONAL verdict, since the issue is fixable through source-level reconciliation and editorial qualification. I therefore recommend no change to the reader's verdict.","tokens_in":25185,"tokens_out":6384,"duration_ms":55568,"concrete_test":"Obtain the published versions of Ref. [2] (Lin et al., PRL 135, 063401, 2025) and the 99.5% two-qubit fidelity result (Evered et al., Nature 622, 268, 2023) and compare the exact fidelity definitions: gate type, measurement protocol (randomized benchmarking, interleaved RB, or full process tomography), qubit subset size, and whether post-selection was applied. Then reconstruct Table 1's 'Max 2Q fidelity' row using one common metric. If the 99.84% and 99.5% numbers are not comparable, the review must either qualify the 99.84% entry or update §§5.1 and 6.1 to the higher baseline; if they are comparable and valid, the manuscript's persistent use of 99.5% as the current ceiling is internally contradicted and needs correction before the bottleneck conclusions can stand.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central map of the field is carried by quantitative fidelity milestones, but the manuscript contains two incompatible baselines for current two-qubit-gate fidelity. Subsection 4.3 and Table 1 report a 'Max 2Q fidelity' of 99.84% (Ref. [2], USTC 2,024-atom array). Subsections 5.1 and 6.1, however, repeatedly build the bottleneck analysis on a 'current 99.5% level'—roadmaps 'assume that gate fidelity can be improved from the current 99.5% level to above 99.9%' and §6.1 states that 'the current two-qubit gate fidelity of 99.5% has exceeded' the surface-code threshold. The concluding trend summary in §5.2 even says fidelity 'increased from the 97% level to 99.84%.' If 99.84% is a genuine, comparably measured number at scale, then the 99.5% baseline is stale and the claim that breaking above 99.9% requires new physics is not supported by the review's own data—the gap is already 0.06 percentage points. If 99.84% is instead a best-case metric measured under different conditions (different gate definition, post-selection, or a subset of qubits) than the 99.5% value used for threshold comparison, then the Table 1 'Max 2Q fidelity' entry and the 'internationally leading' comparisons are overstated. Either way, the central quantitative comparison cannot be evaluated, and the scalability–fidelity bottleneck analysis in §6.1 and the roadmap risk commentary in §5.1 rest on an unresolved internal inconsistency.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript is an English-language review of neutral-atom quantum computing, covering physical principles (qubit encoding, optical trapping, Rydberg interactions and blockade gates, rearrangement), the three main technical routes, 2000–2026 progress milestones, the domestic and international industrialization landscape, and a bottleneck analysis. Its central claim is that optical-tweezer arrays combined with Rydberg interactions are the mainstream route and are progressing toward practical fault-tolerant quantum computation, supported by cited milestones such as roughly 11,000 trapped atoms, a 6,100-atom coherent array, two-qubit gate fidelity of 99.84%, logical-qubit demonstrations, toric-code error correction, and a 2:1 memory-only encoding rate.","tokens_in":25641,"tokens_out":8593,"duration_ms":69792,"significance":"If the cited milestones are reliable, the review is a timely and useful map of the field: it organizes the 2021–2026 developments into physical-basis, hardware-scale, error-correction, and application strands; it explicitly flags in §4.3 and §6.2 that the 2:1 encoding rate is verified only for memory, not computation; and its §6 bottleneck taxonomy offers a structured framework beyond a simple literature list. These features make it potentially valuable as a reference for newcomers and for strategic assessments. The value is conditional, however, on reconciling the fidelity baselines used for threshold and roadmap analysis, on separating trapped atoms from demonstrated coherent qubits, and on consistently grading peer-reviewed results against preprints and company announcements.","major_comments":[{"comment":"The manuscript uses two incompatible baselines for current two-qubit-gate fidelity. Section 4.3 and Table 1 report a maximum two-qubit fidelity of 99.84% (Ref. [2], USTC 2024-atom array), and §5.2 states that fidelity increased from the 97% level to 99.84%. However, §5.1's roadmap commentary and §6.1 repeatedly treat 'the current 99.5% level' as the baseline: §5.1 says roadmaps assume improvement 'from the current 99.5% level to above 99.9%,' and §6.1 states that 'the current two-qubit gate fidelity of 99.5%' has exceeded the surface-code threshold. If 99.84% is a genuine, comparably measured value at scale, then the 99.5% baseline is stale and the claim that reaching 99.9% requires new physical mechanisms is not supported by the paper's own data. If, instead, 99.84% was obtained under different conditions (e.g., a different gate definition or measurement protocol), then Table 1's 'Max 2Q fidelity' entry and the 'internationally leading' comparisons in §5.2 require qualification. The authors must reconcile these figures or explicitly state which values are directly comparable for threshold analysis.","section":"§4.3, Table 1 vs. §5.1, §6.1"},{"comment":"The 11,000-atom metasurface result (Ref. [63], arXiv:2606.02715) is described as 'marking the first time that usable qubit resources in any quantum computing platform reached the ten-thousand level.' The cited work reports stable trapping and rearrangement of about 11,022 atoms; no single-qubit coherence time, gate fidelity, or coherent manipulation at that scale is cited in the manuscript. Calling trapped atoms 'usable qubit resources' overstates the milestone. The text and Table 1 should distinguish 'atoms trapped' from 'qubits with demonstrated coherent control,' as is effectively done for the 6,100-atom Caltech array in Ref. [40].","section":"§4.3(2), Table 1"},{"comment":"Several quantitative milestones that anchor the survey are drawn from non-peer-reviewed sources and are presented with the same confidence as peer-reviewed results: the 11,000-atom trapping claim (Ref. [63], arXiv preprint), the Atom Computing toric-code demonstration (Ref. [30], company press release), the Infleqtion breakthroughs (Ref. [37], press release), and the Hanyuan 2 specifications (Ref. [69], company website). Since these feed Table 1 and the §4.3/§5 progress narrative, the manuscript should label each as 'announced' or 'preprint' at the point of use, or restrict quantitative tables to peer-reviewed results. Without such source grading, readers cannot assess the reliability of the headline milestones.","section":"Table 1, §4.3, §5.1"}],"minor_comments":[{"comment":"The headline phrase 'encoding rates exceeding 1/2' does not carry the memory-only caveat that the body carefully places in §4.3 and §6.2; the abstract should state that this rate applies to error-correction memory, not to computation with logical gates.","section":"Abstract, Innovation Statement"},{"comment":"The phrase 'the theoretical threshold of surface-code quantum error correction (about 1%)' should specify 'error rate of about 1%' to avoid confusing the error-rate threshold with a fidelity threshold; §3.1 states this correctly, so the wording in §2.4 is inconsistent.","section":"§2.4"},{"comment":"The trap-lifetime entry '6,000 s [24]' for the tweezer-and-Rydberg route should be footnoted as a cryogenic-environment result; room-temperature tweezer lifetimes are far shorter, and the table as printed implies a general capability.","section":"Table 1"},{"comment":"The 19 milestones shown in the serpentine timeline are not enumerated in the text; adding a legend or numbered list would allow readers to know which events correspond to which timeline nodes.","section":"Figure 7"},{"comment":"The term 'Förster resonance gates' is introduced without definition; a one-sentence explanation or a pointer to Ref. [55] would help non-specialist readers understand why this route is proposed as a path beyond the 99.9% fidelity barrier.","section":"§5.1"}],"recommendation":"major_revision","confidential_remarks":"The paper draws on a number of 2026 preprints and company announcements, which is understandable for a fast-moving field but should be transparent in the published version. There is also noticeable reliance on the authors' institutional ecosystem in §5.2, including a self-citation in Ref. [51]; this is not disqualifying for a domestic-progress section, but the authors should ensure that self-reported domestic results receive the same source grading as international ones. If the journal does not normally publish surveys, the editor may want to weigh the review's scope against the journal's preferences."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a genuinely useful survey of neutral-atom quantum computing from 2000 through mid-2026, and the authors do several things well. It is not a research paper and does not claim to be. The real value is the synthesis: the three-route comparison, the staged timeline, the identification of error-correction-memory versus error-correction-computation as the key gap, and the candid note that the 2:1 encoding rate is memory-only. The domestic/international coverage is a plus, even if it is uneven.\n\nThe main thing to fix before publication is the fidelity-baseline inconsistency. Table 1 and Section 4.3 report 99.84% as the max two-qubit gate fidelity (USTC, 2024-atom array). But Sections 5.1 and 6.1 repeatedly anchor the bottleneck analysis on \"the current 99.5% level,\" and Section 6.1 says the current fidelity has exceeded the surface-code threshold but still needs to reach 99.9%. If 99.84% is comparable, then the 99.5% baseline is stale, and the claim that breaking 99.9% requires new physics is not supported by the review's own data—the gap is 0.06 points, not 0.4. If the 99.84% was measured under different conditions (smaller scale, post-selection, different gate definition), the review must say that explicitly and redo the bottleneck discussion. Either way, the quantitative comparisons cannot be evaluated as written.\n\nTwo smaller issues. The abstract says \"encoding rates exceeding 1/2\" without the memory-only caveat that the body states clearly; the abstract should carry the same qualifier. And the heavy reliance on company press releases and arXiv preprints for key numbers (11,000 atoms, toric code, Infleqtion) is defensible for a review but should be labeled as such, with a note on verification status. A few \"internationally leading\" claims would benefit from a more explicit non-Chinese benchmark comparison, but that is a style point.\n\nNone of this is fatal. The paper's central map—neutral atoms with tweezers plus Rydberg gates are advancing fast, with QEC moving from memory to computation as the key challenge—holds up. The review deserves a serious referee, and after the fidelity baseline is reconciled it would be a reliable reference for newcomers, funders, and researchers wanting a current landscape map. I would cite it. Recommend accept after major revision, primarily to resolve the 99.5/99.84 inconsistency and qualify the abstract.","headline":"A useful, current survey of neutral-atom QC whose bottleneck analysis rests on two inconsistent fidelity baselines—fix that and the quantitative map becomes trustworthy.","tokens_in":26043,"tokens_out":2403,"would_cite":true,"duration_ms":19977,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Neutral-atom quantum computing has reached 11,000-atom arrays, 99.84% gate fidelity, and first logical processors; the review argues it is now the leading route to practical fault tolerance.","keywords":["neutral atom quantum computing","Rydberg blockade","optical tweezer arrays","quantum error correction","fault-tolerant quantum computing","atom rearrangement","metasurface optical tweezers"],"falsifier":"An independent reproduction or audit of the 2026 metasurface experiment that fails to trap on the order of 10,000 atoms stably, or a direct test showing that the 2:1 encoding rate cannot be maintained once logical gates are executed, would undermine the review's central trajectory claim. A simpler check: if two-qubit fidelity cannot be pushed past 99.9 percent without Förster-resonance gates, then the roadmap assumption of low-overhead fault tolerance by 2027 fails.","tokens_in":24988,"feed_emoji":"⚛️","tokens_out":5252,"duration_ms":43878,"temperature":0.7,"pith_summary":"This review tries to establish that neutral-atom quantum computing, based on laser-trapped atoms and Rydberg interactions, has become the most promising hardware route toward practical fault-tolerant quantum computing. It assembles a milestone record from 2000 to 2026: two-qubit gate fidelity rising from roughly 58 percent in 2010 to 99.84 percent, arrays reaching about 11,000 trapped atoms, continuous operation of a 3,000-qubit system, and demonstrations of logical qubits, toric-code error correction, and a 2:1 memory-encoding rate. The paper argues that the platform's reconfigurable connectivity and room-temperature operation give it structural advantages over superconducting and trapped-ion computers, while identifying the scalability-fidelity trade-off, error-correction computation, atom loss, and laser industrialization as the bottlenecks that decide whether those advantages pay off. A careful reader should care because the review's quantitative map determines which near-term quantum computing milestones are realistic and which are marketing projections.","feed_headline":"Neutral atoms reach 11,000 qubits and near-fault-tolerant error correction","feed_subtitle":"A field review maps the route: 99.84% gate fidelity, logical qubits, and the bottlenecks still blocking practical quantum computers.","key_machinery":"The Rydberg blockade is the physical engine of neutral-atom computing: when one atom is excited to a high-lying Rydberg state, its strong dipole-dipole or van der Waals interaction shifts the Rydberg level of any atom within a blockade radius, so at most one atom in that region can be excited. A three-pulse sequence on control and target atoms converts this blockade into a controlled-phase gate, and combined with single-qubit rotations it gives a universal gate set. The second load-bearing piece is the reconfigurable optical tweezer array, whose move-entangle-separate operation gives all-to-all connectivity and lets the platform host surface codes, LDPC codes, and zoned architectures on the same hardware. Together these two mechanisms carry the review's claim that neutral atoms can combine scale, connectivity, and gate fidelity in one machine.","core_discovery":"The central claim is that optical tweezer arrays combined with Rydberg blockade now constitute the mainstream and most advanced technical route in quantum computing hardware. The paper documents a chain of results: random arrays of atoms are rearranged into defect-free, reconfigurable configurations; Rydberg blockade turns pairs into high-fidelity entangling gates; and the same platform has demonstrated fault-tolerant primitives from transversal logical gates to loss-aware decoding. In the review's own framing, the milestone numbers — 11,022 atoms in a metasurface tweezer array, a 99.84 percent two-qubit gate, 48 logical qubits, six rounds of toric-code error correction, and an encoding rate exceeding 1/2 for quantum memory — show that neutral atoms are scaling faster than other platforms. The paper is equally explicit about the line not yet crossed: the 2:1 encoding rate was verified only for storage, and error-correction computation under logical gates remains unproven.","pith_inferences":["The paper's numbers imply that qubit count is no longer the differentiator between platforms; the next competitive axis is error-correction computation and gate fidelity above 99.9 percent, where the review itself notes new physical mechanisms may be needed.","A testable extension follows from the metasurface result: if a single metasurface can replace a microscope objective for array generation, the cost per qubit falls sharply, and the practical limit of the platform shifts to laser power and control electronics rather than optics.","The review's own risk commentary suggests that company roadmaps to 100 logical qubits by 2026-2029 are optimistic estimates; a reader should treat those dates as upper bounds until the memory-to-computation step is experimentally demonstrated.","An implicit consequence is that if Förster-resonance gates push fidelity above 99.9 percent, the neutral-atom platform could close the fidelity gap to trapped ions while keeping its scaling edge, making it the most versatile quantum computing platform in the near term."],"forward_implications":["If the reported milestone numbers hold, the neutral-atom platform should reach ten-thousand-qubit physical processors and hundred-logical-qubit demonstrations within the paper's five-to-ten-year window.","Exceeding the surface-code threshold is not enough on its own: measurement, atom loss, and feedback must also meet threshold conditions before fault-tolerant computation becomes practical.","The 2:1 encoding rate, if later extended from memory to computation, would cut the physical-qubit cost of fault tolerance by orders of magnitude relative to surface codes.","The move-entangle-separate architecture makes high-rate LDPC codes a natural fit, so error-correction overhead on neutral atoms may end up lower than on fixed-topology platforms.","Bottlenecks in control electronics and compilers, not just physics, will determine whether arrays above ten thousand atoms can be operated with full parallelism."],"supporting_citations":[{"why":"Supplies the original theoretical proposal for fast quantum gates using Rydberg dipole interactions and blockade.","marker":"[10]"},{"why":"Provides the quantitative feasibility analysis and decoherence budget that set experimental design benchmarks.","marker":"[11]"},{"why":"Foundational review of quantum information with Rydberg atoms that frames the field's physical principles.","marker":"[12]"},{"why":"Reports the first programmable logical quantum processor, the central evidence that fault tolerance moved from theory to practice.","marker":"[1]"},{"why":"Demonstrates high-fidelity parallel entangling gates on a neutral-atom computer, the fidelity baseline for the review's comparisons.","marker":"[3]"},{"why":"Reports AI-enabled parallel assembly of 2024 defect-free atoms with 99.84% two-qubit gate fidelity, a key scaling and fidelity milestone.","marker":"[2]"},{"why":"Documents a 6100-atom coherent cesium tweezer array, evidence for thousand-qubit-scale qubit resources.","marker":"[40]"},{"why":"Shows continuous operation of a 3000-qubit system with fast atom replenishment, the engineering path against atom loss.","marker":"[41]"},{"why":"Reports stable trapping of about 11,000 atoms with a single metasurface, the current scale record the review builds on.","marker":"[63]"},{"why":"Proposes a fault-tolerant neutral-atom architecture with loss-aware decoding, the framework the review presents as the route ahead.","marker":"[27]"}],"fun_headline_variants":["Neutral atoms hit 11k qubits, but fault tolerance still elusive","Scaling neutral atoms: 11k qubits, but error correction lags","11k qubits in neutral atoms, but logical gates still unproven","Neutral atoms scale fast, but fault-tolerant computing remains distant"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The review's thesis depends on the accuracy of reported milestones taken from secondary sources, especially the 11,000-atom array, the 99.84 percent two-qubit fidelity, and the 2:1 memory encoding rate; if any of these numbers is later corrected or shown to apply only under narrow conditions, the cross-platform comparisons and trend statements would need revision.","fun_headline_variants_meta":{"raw":{"variants":["Neutral atoms hit 11k qubits, but fault tolerance still elusive","Scaling neutral atoms: 11k qubits, but error correction lags","11k qubits in neutral atoms, but logical gates still unproven","Neutral atoms scale fast, but fault-tolerant computing remains distant"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000662,"raw_usage":{"total_tokens":3059,"prompt_tokens":1010,"completion_tokens":2049,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":1968}},"tokens_in":626,"tokens_out":2049,"duration_ms":13746,"temperature":1.0,"reasoning_tokens":1968,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T04:12:51.569007+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An independent reproduction or audit of the 2026 metasurface experiment that fails to trap on the order of 10,000 atoms stably, or a direct test showing that the 2:1 encoding rate cannot be maintained once logical gates are executed, would undermine the review's central trajectory claim. A simpler check: if two-qubit fidelity cannot be pushed past 99.9 percent without Förster-resonance gates, then the roadmap assumption of low-overhead fault tolerance by 2027 fails.","supporting_citations":[{"cited_title":"Fast quan- tum gates for neutral atoms[J].Physical Review Letters, 2000, 85: 2208","cited_arxiv_id":null,"evidence_quote":"Supplies the original theoretical proposal for fast quantum gates using Rydberg dipole interactions and blockade."},{"cited_title":"Analysis of a quantum logic device based on dipole-dipole interactions of optically trapped Rydberg atoms[J].Physical Review A, 2005, 72: 022347","cited_arxiv_id":null,"evidence_quote":"Provides the quantitative feasibility analysis and decoherence budget that set experimental design benchmarks."},{"cited_title":"Quantum information with Rydberg atoms[J].Reviews of Modern Physics, 2010, 82: 2313–2363","cited_arxiv_id":null,"evidence_quote":"Foundational review of quantum information with Rydberg atoms that frames the field's physical principles."},{"cited_title":"High-fidelity parallel entangling gates on a neutral-atom quantum computer[J].Nature, 2023, 622: 268–272","cited_arxiv_id":null,"evidence_quote":"Demonstrates high-fidelity parallel entangling gates on a neutral-atom computer, the fidelity baseline for the review's comparisons."},{"cited_title":"Trapping 11,000 Atoms in a Tweezer Array Generated by a Single Metasurface","cited_arxiv_id":"2606.02715","evidence_quote":"Reports stable trapping of about 11,000 atoms with a single metasurface, the current scale record the review builds on."},{"cited_title":"A fault-tolerant neutral-atom architecture for uni- versal quantum computation[J].Nature, 2026, 649: 39–46","cited_arxiv_id":null,"evidence_quote":"Proposes a fault-tolerant neutral-atom architecture with loss-aware decoding, the framework the review presents as the route ahead."}],"review_version":2}