Tech # Outlook for Quantum Computers: What’s Likely Next (2026–2035+) # 양자컴퓨터의 향후 전망…
Page Info
Writer Joshuaa
Hit 414 Hits
Date 26-01-01 22:14
Content
# Outlook for Quantum Computers: What’s Likely Next (2026–2035+)
# 양자컴퓨터의 향후 전망: 2026–2035+ 로드맵과 현실적 시나리오
---
## English
### 1) Where quantum computing really is today (as of early 2026)
Quantum computing is moving from the “NISQ” phase (Noisy Intermediate-Scale Quantum—useful experiments but error-prone) toward the first **fault-tolerant building blocks**, driven mainly by **quantum error correction (QEC)** progress.
Two highly visible signals of that shift:
* **Google (Dec 2024)** reported surface-code scaling where the encoded error rate improves as the code gets larger (“exponential” improvement with increasing code distance on its Willow processor). ([Google Research][1])
* **IBM (Jun 10, 2025)** published an end-to-end framework and updated roadmap aimed at **large-scale fault tolerance by 2029** (Starling), explicitly framing the journey as “modules + QEC + real-time decoding.” ([IBM][2])
The bottom line: the field is past “pure demo” and into “engineering the stack,” but **useful, general-purpose fault-tolerant quantum computers** remain a **late-2020s/2030s** target in the most credible roadmaps.
---
### 2) The most likely timeline: 3 horizons that matter
#### Horizon A: 2026–2028 — “Hybrid advantage” and narrow wins
Expect more results where quantum is used as an **accelerator alongside HPC** (high-performance computing), but only for **carefully chosen problems**.
IBM, for example, states confidence that users will deliver “quantum advantage” **by the end of 2026** with quantum serving as an accelerator for classical HPC, and its roadmap targets early “scientific quantum advantage” examples in 2026. ([IBM][2])
What this will look like in practice:
* Chemistry / materials workflows where a quantum subroutine reduces a bottleneck in a classical pipeline
* Optimization heuristics where quantum provides “one more tool,” not a universal replacement
* More credible benchmarking culture (because governments are funding “quantum benchmarking” efforts at scale) ([The Department of Energy's Energy.gov][3])
#### Horizon B: 2029–early 2030s — First “real” fault-tolerant machines, still scarce
Roadmaps cluster around the **end of the 2020s** for early fault-tolerant systems:
* IBM targets **Starling (2029)**: ~**200 logical qubits** and the ability to run circuits with **100 million quantum gates**. ([IBM][2])
* Quantinuum publicly targets a universal, fault-tolerant system “by 2030.” ([Quantinuum][4])
Key reality: “fault-tolerant exists” does not mean “widely accessible.” Early FT machines will be expensive, capacity-limited, and heavily booked—much like the earliest supercomputers.
#### Horizon C: mid-2030s+ — Scaling logical qubits and expanding practical domains
If QEC overheads fall and manufacturing scales, you get:
* more logical qubits,
* deeper circuits,
* and broader application portfolios (materials, pharma, cryptography-adjacent tasks, niche ML).
But this depends on achieving **repeatable, manufacturable** performance—not just one lab’s best device.
---
### 3) Technology paths: who’s betting on what (and why that matters)
Quantum computing is not one technology. Different hardware approaches trade off speed, fidelity, connectivity, manufacturability, and scaling risk.
**(1) Superconducting qubits (IBM, Google, others)**
Pros: fast gates, strong ecosystem, rapid iteration.
Cons: wiring/cryogenics complexity at scale; QEC overhead is large until logical qubits are robust.
Recent QEC headlines are largely in this camp (Google’s Willow QEC scaling; IBM’s modular FT plan). ([Google Research][1])
**(2) Trapped ions (IonQ, Quantinuum)**
Pros: very high-quality qubits; long coherence; flexible connectivity.
Cons: gate speeds and scaling engineering differ; packaging and multiplexing are hard.
IonQ publishes an aggressive long-range roadmap (e.g., very large physical and logical-qubit targets by 2030). ([IonQ][5])
**(3) Photonics (PsiQuantum and others)**
Pros: leverages semiconductor fab + telecom components; potentially good for modular scaling.
Cons: building a full fault-tolerant photonic stack is extremely demanding.
Large capital commitments and site builds are underway (e.g., PsiQuantum facility announcements). ([Reuters][6])
**(4) Topological qubits (Microsoft’s bet)**
Pros (if it works): hardware-level error suppression could reduce QEC overhead dramatically.
Cons: hardest scientific/engineering validation path.
Microsoft describes topological-qubit milestones and announced “Majorana 1” as a step on that roadmap. ([Microsoft Azure][7])
**(5) Quantum annealing (D-Wave)**
Pros: commercially offered systems today for certain optimization-style workloads.
Cons: not a universal gate-model machine; “advantage” claims depend heavily on problem framing.
D-Wave markets Advantage2 for real-time scalable optimization, and it has reported commercial system sales. ([dwavequantum.com][8])
---
### 4) What quantum computers will be good for (and what they won’t)
#### Strongest long-term candidates
1. **Chemistry & materials simulation**
This is the most frequently cited “killer app” because quantum systems naturally model quantum matter. Google explicitly highlights chemistry/drug discovery as a key target domain. ([Google Research][1])
2. **Cryptography-related impacts (mostly indirect in the short run)**
The *main* near-term consequence is not “quantum breaks everything tomorrow,” but that security teams must plan for migration due to the **harvest-now, decrypt-later** risk.
NIST finalized the first set of Post-Quantum Cryptography standards in **August 2024** (FIPS 203 ML-KEM, FIPS 204 ML-DSA, FIPS 205 SLH-DSA) and explicitly states these can and should be put into use now. ([NIST][9])
3. **Optimization and sampling (selective)**
Quantum may produce speedups for certain structured problems, but broad “optimization supremacy” is unlikely. Many problems have excellent classical heuristics, and proving quantum advantage is difficult.
#### What quantum will *not* do soon
* Replace GPUs/CPUs for mainstream AI training
* Magically solve NP-hard problems at arbitrary scale
* Deliver consumer-scale “desktop quantum computers” anytime soon (cryogenics, control, and error correction make that unrealistic)
---
### 5) Macro outlook: investment, geopolitics, and “why 2025 was a turning point for attention”
Quantum is now treated as strategic national infrastructure:
* The UN proclaimed **2025 as the International Year of Quantum Science and Technology**, reflecting global policy attention and workforce-building emphasis. ([UN Docs][10])
* The U.S. DOE announced **$625 million** (Nov 4, 2025) to renew five National QIS Research Centers. ([The Department of Energy's Energy.gov][3])
* The EU Quantum Technologies Flagship is framed as a long-term initiative with an expected **€1 billion** EU budget. ([디지털 전략][11])
This matters because the limiting factors are increasingly “industrial”:
* fabrication yield and packaging,
* cryogenic supply chains,
* control electronics,
* datacenter-grade power/operations,
* and talent at the intersection of physics and systems engineering.
---
### 6) Practical guidance that holds up under uncertainty
For organizations planning around quantum (technology, security, or R&D), the most robust approach is:
1. **Separate “NISQ pilots” from “fault-tolerant business cases.”**
Treat near-term results as learning and capability-building, not guaranteed ROI.
2. **Adopt post-quantum cryptography on a real schedule.**
Inventory cryptographic dependencies; prioritize long-lived secrets; deploy PQC/hybrid where feasible, aligned with NIST’s finalized standards. ([NIST][9])
3. **Build hybrid workflows now.**
The credible roadmaps emphasize quantum + HPC integration, not quantum alone. ([IBM][12])
4. **Demand benchmarks tied to your data and constraints.**
Quantum “advantage” is rarely universal; it’s a property of a workload + cost model + error budget.
---
## 한국어
### 1) 2026년 초 기준, 양자컴퓨터의 ‘현실 위치’
양자컴퓨터는 여전히 **NISQ(노이즈가 큰 중간규모)** 단계의 한계를 갖고 있지만, 최근 흐름은 “쇼케이스 실험”을 넘어 **오류정정(QEC) 기반의 공학적 스택 구축**으로 확실히 이동했습니다.
대표적 신호:
* **구글(2024년 12월)**: 표면코드 기반에서 코드 크기를 키울수록 논리 큐비트의 오류가 더 잘 억제되는(스케일 업에 따라 성능이 “지수적으로” 좋아지는) 결과를 Willow 프로세서로 보고. ([Google Research][1])
* **IBM(2025년 6월 10일)**: 2029년 대규모 내결함성(FTQC)을 목표로, 모듈형 아키텍처 + QEC + 실시간 디코딩을 포함한 “엔드투엔드” 프레임워크와 로드맵을 공개. ([IBM][2])
정리하면, “아직 늦었다”도 “당장 된다”도 아니고, **실용 FTQC를 향한 엔지니어링 레이스가 시작된 단계**입니다.
---
### 2) 가장 그럴듯한 시간표: 3개의 구간으로 보는 전망
#### A구간: 2026–2028 — “하이브리드(양자+HPC) 이점”의 등장
이 시기에는 양자가 단독으로 승리하기보다 **HPC 파이프라인의 가속기**로서 특정 문제에서 의미 있는 결과가 나올 가능성이 큽니다.
IBM은 **2026년 말까지** 양자를 고전 HPC의 가속기로 활용하는 형태의 “양자 이점”을 기대한다고 밝히고, 2026년에 초기 “과학적 양자 이점” 사례를 목표로 합니다. ([IBM][2])
실제 형태는 대체로:
* 화학/소재에서 특정 서브루틴을 양자가 맡아 병목을 줄이는 방식
* 최적화에서 “양자+고전” 혼합 휴리스틱
* 정부 차원의 벤치마킹/검증 프로그램 확대(대규모 투자와 평가 체계) ([The Department of Energy's Energy.gov][3])
#### B구간: 2029–2030년대 초 — 최초의 ‘진짜’ 내결함성 시스템(희소 자원)
로드맵이 모이는 지점이 **2029 전후**입니다.
* IBM: **2029년 Starling**로 **200 논리 큐비트**, **1억(100M) 게이트** 규모 회로 실행을 제시. ([IBM][2])
* Quantinuum: **2030년까지** 범용 내결함성 시스템 목표를 공표. ([Quantinuum][4])
다만 이때도 “모든 기업이 바로 쓰는 시대”가 아니라, 초기 FT 장비는 **비싸고, 희소하고, 예약이 꽉 찬** 형태가 될 가능성이 높습니다.
#### C구간: 2030년대 중반+ — 논리 큐비트 확장과 응용 영역 확대
QEC 오버헤드가 낮아지고 제조·운영이 산업화되면:
* 더 많은 논리 큐비트,
* 더 깊은 회로,
* 더 다양한 산업 응용(소재, 신약, 특정 형태의 최적화/샘플링 등)
으로 이어질 수 있습니다.
---
### 3) 기술 노선이 여러 개인 이유(그리고 전망에 미치는 영향)
* **초전도(IBM/Google 등)**: 빠른 게이트와 생태계 강점. 대신 대규모 배선/극저온/오류정정 비용이 관건. ([IBM][2])
* **이온 트랩(IonQ/Quantinuum)**: 큐비트 품질이 강점. 다만 확장 공학(패키징/멀티플렉싱)이 난제. IonQ는 2030년을 향해 공격적 로드맵을 제시합니다. ([IonQ][5])
* **포토닉(PsiQuantum 등)**: 반도체 공정과 광통신 기반 확장을 노리지만, FT 스택 구축 난이도가 매우 큼. 대규모 시설/투자가 진행 중입니다. ([Reuters][6])
* **위상학(Microsoft)**: 성공하면 하드웨어 레벨에서 오류를 줄여 QEC 부담을 크게 낮출 잠재력. 대신 검증 난이도가 가장 높은 축. Microsoft는 Majorana 1을 로드맵의 단계로 제시합니다. ([Microsoft Azure][7])
* **어닐링(D-Wave)**: 특정 최적화 스타일 업무에 상용 제공을 강조. 범용 게이트형과 성격이 다르고 “이점”은 문제 정의에 크게 의존. ([dwavequantum.com][8])
---
### 4) 어디에 쓸모가 커지나 / 무엇은 당분간 어렵나
**유망도가 가장 높은 축**
1. **화학·소재 시뮬레이션**: 양자계는 양자계로 모사할 때 근본적 이점이 기대됩니다. 구글도 화학·신약을 대표 응용으로 언급합니다. ([Google Research][1])
2. **보안(암호) 영향: “지금 당장 붕괴”보다 “이행이 시급”**
중요한 건 ‘오늘 밤 해킹’이 아니라 **수년 뒤 복호화될 위험(수집 후 추후 복호화)** 때문에 장기 비밀이 위험해진다는 점입니다.
NIST는 **2024년 8월** PQC 표준(ML-KEM/ML-DSA/SLH-DSA)을 확정했고, “지금 사용해도 된다/해야 한다”는 메시지를 명확히 냅니다. ([NIST][9])
3. **최적화/샘플링(선별적)**: 넓게 다 이긴다기보다 “특정 구조의 문제”에서만 승부가 납니다.
**당분간 어려운 것**
* GPU/CPU를 대체해 주류 AI 학습을 맡는 것
* NP-hard 문제를 만능으로 대규모 해결
* 소비자용 ‘데스크톱 양자컴퓨터’(극저온/제어/오류정정이 본질적으로 무겁습니다)
---
### 5) 거시 전망: 국가 전략 인프라로 굳어지는 중
* UN은 **2025년을 ‘세계 양자과학·기술의 해’**로 선포했습니다. ([UN Docs][10])
* 미국 DOE는 **2025년 11월 4일** 국가 QIS 연구센터 5곳에 **6억 2,500만 달러** 재투자를 발표했습니다. ([The Department of Energy's Energy.gov][3])
* EU는 양자 플래그십을 장기 R&D 이니셔티브로 두고 **EU 예산 €10억** 규모를 명시합니다. ([디지털 전략][11])
이제 핵심 병목은 연구만이 아니라, **제조 수율·패키징·극저온 공급망·제어 전자·전력/데이터센터 운영·인재** 같은 산업 요소입니다.
---
### 6) 불확실성을 견디는 실무 전략(추가 팁·응용)
1. **NISQ 파일럿과 FTQC 비즈니스 케이스를 분리**해서 관리해야 합니다.
2. **PQC 이행은 “계획”이 아니라 “프로젝트”**가 되어야 합니다(암호 인벤토리, 장기 비밀 우선, 하이브리드 전환). ([NIST][9])
3. 로드맵의 주류는 “양자 단독”이 아니라 **양자+HPC 결합**입니다. ([IBM][12])
4. “양자 이점”은 대부분 **특정 워크로드·비용모델·오류예산**에 종속됩니다. 따라서 벤치마크는 반드시 **자사 제약조건**과 연결되어야 합니다.
---
## 日本語
### 1) 2026年初頭の到達点
量子計算は依然としてノイズの大きい段階（NISQ）にありますが、焦点は「デモ」から **誤り訂正（QEC）に基づく工学化**へ移りました。
GoogleはWillowで、表面コードの距離を増やすほど符号化誤りが抑制される（スケールで改善する）結果を報告しています。 ([Google Research][1])
IBMは2029年の大規模フォールトトレラント（Starling）を明確に掲げ、モジュール化と実時間デコーディングを含む枠組みを提示しています。 ([IBM][2])
### 2) 時間軸（現実的な3区分）
* **2026–2028**：HPCと組み合わせた“狭い優位（ハイブリッド優位）”が中心。IBMは2026年末までの優位の可能性を述べています。 ([IBM][2])
* **2029〜2030年代初頭**：初期の本格FT機が出るが希少・高価。IBMは2029年に200論理量子ビット/1億ゲート級を掲げています。 ([IBM][2])
* **2030年代中盤以降**：論理量子ビットの規模拡大と応用領域の広がり（ただし産業化が前提）。
### 3) “用途”の見立て（強い順）
* **化学・材料**：量子系を量子で扱う必然性があり有望。 ([Google Research][1])
* **暗号（移行の緊急性）**：NISTは2024年8月にPQC標準（ML-KEM/ML-DSA/SLH-DSA）を確定。 ([NIST][9])
* **最適化**：万能ではなく、問題構造次第。
### 4) 戦略（不確実性に強い）
PQC移行を本格プロジェクト化し、NISQの実験とFTの事業化を切り分け、ハイブリッド（量子+HPC）前提でワークフローを構築するのが最も堅い方針です。 ([IBM][12])
---
## Español
### 1) Estado real (inicio de 2026)
La disciplina está pasando de “demos ruidosas” a **ingeniería con corrección de errores (QEC)**.
Google reportó mejoras de QEC al aumentar el tamaño del código (Willow). ([Google Research][1])
IBM publicó un marco y una hoja de ruta hacia **tolerancia a fallos a gran escala en 2029** (Starling). ([IBM][2])
### 2) Pronóstico por horizontes
* **2026–2028**: “ventajas híbridas” (cuántica + HPC) en casos estrechos; IBM apunta a ejemplos de ventaja alrededor de 2026. ([IBM][2])
* **2029–principios de los 2030**: primeras máquinas tolerantes a fallos, pero escasas; IBM menciona 200 qubits lógicos y 100M compuertas para 2029. ([IBM][2])
* **2030s+**: ampliación de qubits lógicos y portafolio de aplicaciones si la industrialización reduce el overhead.
### 3) Impacto en seguridad (lo más accionable hoy)
NIST finalizó en **agosto de 2024** los primeros estándares de criptografía post-cuántica (ML-KEM, ML-DSA, SLH-DSA) y señala que deben adoptarse ya. ([NIST][9])
---
## Français
### 1) Situation (début 2026)
Le secteur bascule de la phase NISQ vers une **industrialisation guidée par la correction d’erreurs (QEC)**.
Google a présenté des résultats de QEC où l’encodage s’améliore en grandissant (Willow). ([Google Research][1])
IBM vise un jalon clair de **tolérance aux fautes à grande échelle en 2029** (Starling). ([IBM][2])
### 2) Horizons plausibles
* **2026–2028** : gains “hybrides” (quantique + HPC) sur des problèmes étroits ; IBM évoque des exemples d’avantage autour de 2026. ([IBM][2])
* **2029–début 2030s** : premières machines FT, coûteuses et rares ; IBM annonce 200 qubits logiques et 100M portes pour 2029. ([IBM][2])
* **Milieu 2030s+** : élargissement des cas d’usage si la mise à l’échelle devient manufacturable.
### 3) Le point le plus concret dès maintenant : la transition crypto
NIST a finalisé en **août 2024** les premiers standards post-quantiques (ML-KEM, ML-DSA, SLH-DSA) et recommande de les déployer. ([NIST][9])
---
* [Reuters](https://www.reuters.com/business/psiquantum-breaks-ground-chicago-quantum-site-after-1-billion-funding-2025-09-30/?utm_source=chatgpt.com)
* [CT Insider](https://www.ctinsider.com/business/article/quantum-technology-yale-uconn-quantumct-lamont-21197360.php?utm_source=chatgpt.com)
* [statesman.com](https://www.statesman.com/business/technology/article/ut-semiconductor-chips-grant-greg-abbott-21235615.php?utm_source=chatgpt.com)
* [barrons.com](https://www.barrons.com/articles/d-wave-quantum-computing-stock-d8596186?utm_source=chatgpt.com)
[1]: https://research.google/blog/making-quantum-error-correction-work/ "Making quantum error correction work"
[2]: https://www.ibm.com/quantum/blog/large-scale-ftqc "IBM lays out clear path to fault-tolerant quantum computing | IBM Quantum Computing Blog"
[3]: https://www.energy.gov/articles/energy-department-announces-625-million-advance-next-phase-national-quantum-information?utm_source=chatgpt.com "Energy Department Announces $625 Million to Advance ..."
[4]: https://www.quantinuum.com/press-releases/quantinuum-unveils-accelerated-roadmap-to-achieve-universal-fault-tolerant-quantum-computing-by-2030?utm_source=chatgpt.com "Quantinuum Unveils Accelerated Roadmap to Achieve ..."
[5]: https://www.ionq.com/roadmap?utm_source=chatgpt.com "Roadmap"
[6]: https://www.reuters.com/business/psiquantum-breaks-ground-chicago-quantum-site-after-1-billion-funding-2025-09-30/?utm_source=chatgpt.com "PsiQuantum breaks ground on Chicago quantum site after $1 billion funding"
[7]: https://azure.microsoft.com/en-us/blog/quantum/2025/02/19/microsoft-unveils-majorana-1-the-worlds-first-quantum-processor-powered-by-topological-qubits/?utm_source=chatgpt.com "Microsoft unveils Majorana 1, the world's first quantum ..."
[8]: https://www.dwavequantum.com/solutions-and-products/systems/?utm_source=chatgpt.com "The Advantage2™ Quantum Computer"
[9]: https://www.nist.gov/news-events/news/2024/08/nist-releases-first-3-finalized-post-quantum-encryption-standards "NIST Releases First 3 Finalized Post-Quantum Encryption Standards | NIST"
[10]: https://docs.un.org/en/A/RES/78/287?utm_source=chatgpt.com "A/RES/78/287 - General Assembly - the United Nations"
[11]: https://digital-strategy.ec.europa.eu/en/policies/quantum-technologies-flagship?utm_source=chatgpt.com "Quantum Technologies Flagship"
[12]: https://www.ibm.com/roadmaps/quantum/ "www.ibm.com"
# 양자컴퓨터의 향후 전망: 2026–2035+ 로드맵과 현실적 시나리오
---
## English
### 1) Where quantum computing really is today (as of early 2026)
Quantum computing is moving from the “NISQ” phase (Noisy Intermediate-Scale Quantum—useful experiments but error-prone) toward the first **fault-tolerant building blocks**, driven mainly by **quantum error correction (QEC)** progress.
Two highly visible signals of that shift:
* **Google (Dec 2024)** reported surface-code scaling where the encoded error rate improves as the code gets larger (“exponential” improvement with increasing code distance on its Willow processor). ([Google Research][1])
* **IBM (Jun 10, 2025)** published an end-to-end framework and updated roadmap aimed at **large-scale fault tolerance by 2029** (Starling), explicitly framing the journey as “modules + QEC + real-time decoding.” ([IBM][2])
The bottom line: the field is past “pure demo” and into “engineering the stack,” but **useful, general-purpose fault-tolerant quantum computers** remain a **late-2020s/2030s** target in the most credible roadmaps.
---
### 2) The most likely timeline: 3 horizons that matter
#### Horizon A: 2026–2028 — “Hybrid advantage” and narrow wins
Expect more results where quantum is used as an **accelerator alongside HPC** (high-performance computing), but only for **carefully chosen problems**.
IBM, for example, states confidence that users will deliver “quantum advantage” **by the end of 2026** with quantum serving as an accelerator for classical HPC, and its roadmap targets early “scientific quantum advantage” examples in 2026. ([IBM][2])
What this will look like in practice:
* Chemistry / materials workflows where a quantum subroutine reduces a bottleneck in a classical pipeline
* Optimization heuristics where quantum provides “one more tool,” not a universal replacement
* More credible benchmarking culture (because governments are funding “quantum benchmarking” efforts at scale) ([The Department of Energy's Energy.gov][3])
#### Horizon B: 2029–early 2030s — First “real” fault-tolerant machines, still scarce
Roadmaps cluster around the **end of the 2020s** for early fault-tolerant systems:
* IBM targets **Starling (2029)**: ~**200 logical qubits** and the ability to run circuits with **100 million quantum gates**. ([IBM][2])
* Quantinuum publicly targets a universal, fault-tolerant system “by 2030.” ([Quantinuum][4])
Key reality: “fault-tolerant exists” does not mean “widely accessible.” Early FT machines will be expensive, capacity-limited, and heavily booked—much like the earliest supercomputers.
#### Horizon C: mid-2030s+ — Scaling logical qubits and expanding practical domains
If QEC overheads fall and manufacturing scales, you get:
* more logical qubits,
* deeper circuits,
* and broader application portfolios (materials, pharma, cryptography-adjacent tasks, niche ML).
But this depends on achieving **repeatable, manufacturable** performance—not just one lab’s best device.
---
### 3) Technology paths: who’s betting on what (and why that matters)
Quantum computing is not one technology. Different hardware approaches trade off speed, fidelity, connectivity, manufacturability, and scaling risk.
**(1) Superconducting qubits (IBM, Google, others)**
Pros: fast gates, strong ecosystem, rapid iteration.
Cons: wiring/cryogenics complexity at scale; QEC overhead is large until logical qubits are robust.
Recent QEC headlines are largely in this camp (Google’s Willow QEC scaling; IBM’s modular FT plan). ([Google Research][1])
**(2) Trapped ions (IonQ, Quantinuum)**
Pros: very high-quality qubits; long coherence; flexible connectivity.
Cons: gate speeds and scaling engineering differ; packaging and multiplexing are hard.
IonQ publishes an aggressive long-range roadmap (e.g., very large physical and logical-qubit targets by 2030). ([IonQ][5])
**(3) Photonics (PsiQuantum and others)**
Pros: leverages semiconductor fab + telecom components; potentially good for modular scaling.
Cons: building a full fault-tolerant photonic stack is extremely demanding.
Large capital commitments and site builds are underway (e.g., PsiQuantum facility announcements). ([Reuters][6])
**(4) Topological qubits (Microsoft’s bet)**
Pros (if it works): hardware-level error suppression could reduce QEC overhead dramatically.
Cons: hardest scientific/engineering validation path.
Microsoft describes topological-qubit milestones and announced “Majorana 1” as a step on that roadmap. ([Microsoft Azure][7])
**(5) Quantum annealing (D-Wave)**
Pros: commercially offered systems today for certain optimization-style workloads.
Cons: not a universal gate-model machine; “advantage” claims depend heavily on problem framing.
D-Wave markets Advantage2 for real-time scalable optimization, and it has reported commercial system sales. ([dwavequantum.com][8])
---
### 4) What quantum computers will be good for (and what they won’t)
#### Strongest long-term candidates
1. **Chemistry & materials simulation**
This is the most frequently cited “killer app” because quantum systems naturally model quantum matter. Google explicitly highlights chemistry/drug discovery as a key target domain. ([Google Research][1])
2. **Cryptography-related impacts (mostly indirect in the short run)**
The *main* near-term consequence is not “quantum breaks everything tomorrow,” but that security teams must plan for migration due to the **harvest-now, decrypt-later** risk.
NIST finalized the first set of Post-Quantum Cryptography standards in **August 2024** (FIPS 203 ML-KEM, FIPS 204 ML-DSA, FIPS 205 SLH-DSA) and explicitly states these can and should be put into use now. ([NIST][9])
3. **Optimization and sampling (selective)**
Quantum may produce speedups for certain structured problems, but broad “optimization supremacy” is unlikely. Many problems have excellent classical heuristics, and proving quantum advantage is difficult.
#### What quantum will *not* do soon
* Replace GPUs/CPUs for mainstream AI training
* Magically solve NP-hard problems at arbitrary scale
* Deliver consumer-scale “desktop quantum computers” anytime soon (cryogenics, control, and error correction make that unrealistic)
---
### 5) Macro outlook: investment, geopolitics, and “why 2025 was a turning point for attention”
Quantum is now treated as strategic national infrastructure:
* The UN proclaimed **2025 as the International Year of Quantum Science and Technology**, reflecting global policy attention and workforce-building emphasis. ([UN Docs][10])
* The U.S. DOE announced **$625 million** (Nov 4, 2025) to renew five National QIS Research Centers. ([The Department of Energy's Energy.gov][3])
* The EU Quantum Technologies Flagship is framed as a long-term initiative with an expected **€1 billion** EU budget. ([디지털 전략][11])
This matters because the limiting factors are increasingly “industrial”:
* fabrication yield and packaging,
* cryogenic supply chains,
* control electronics,
* datacenter-grade power/operations,
* and talent at the intersection of physics and systems engineering.
---
### 6) Practical guidance that holds up under uncertainty
For organizations planning around quantum (technology, security, or R&D), the most robust approach is:
1. **Separate “NISQ pilots” from “fault-tolerant business cases.”**
Treat near-term results as learning and capability-building, not guaranteed ROI.
2. **Adopt post-quantum cryptography on a real schedule.**
Inventory cryptographic dependencies; prioritize long-lived secrets; deploy PQC/hybrid where feasible, aligned with NIST’s finalized standards. ([NIST][9])
3. **Build hybrid workflows now.**
The credible roadmaps emphasize quantum + HPC integration, not quantum alone. ([IBM][12])
4. **Demand benchmarks tied to your data and constraints.**
Quantum “advantage” is rarely universal; it’s a property of a workload + cost model + error budget.
---
## 한국어
### 1) 2026년 초 기준, 양자컴퓨터의 ‘현실 위치’
양자컴퓨터는 여전히 **NISQ(노이즈가 큰 중간규모)** 단계의 한계를 갖고 있지만, 최근 흐름은 “쇼케이스 실험”을 넘어 **오류정정(QEC) 기반의 공학적 스택 구축**으로 확실히 이동했습니다.
대표적 신호:
* **구글(2024년 12월)**: 표면코드 기반에서 코드 크기를 키울수록 논리 큐비트의 오류가 더 잘 억제되는(스케일 업에 따라 성능이 “지수적으로” 좋아지는) 결과를 Willow 프로세서로 보고. ([Google Research][1])
* **IBM(2025년 6월 10일)**: 2029년 대규모 내결함성(FTQC)을 목표로, 모듈형 아키텍처 + QEC + 실시간 디코딩을 포함한 “엔드투엔드” 프레임워크와 로드맵을 공개. ([IBM][2])
정리하면, “아직 늦었다”도 “당장 된다”도 아니고, **실용 FTQC를 향한 엔지니어링 레이스가 시작된 단계**입니다.
---
### 2) 가장 그럴듯한 시간표: 3개의 구간으로 보는 전망
#### A구간: 2026–2028 — “하이브리드(양자+HPC) 이점”의 등장
이 시기에는 양자가 단독으로 승리하기보다 **HPC 파이프라인의 가속기**로서 특정 문제에서 의미 있는 결과가 나올 가능성이 큽니다.
IBM은 **2026년 말까지** 양자를 고전 HPC의 가속기로 활용하는 형태의 “양자 이점”을 기대한다고 밝히고, 2026년에 초기 “과학적 양자 이점” 사례를 목표로 합니다. ([IBM][2])
실제 형태는 대체로:
* 화학/소재에서 특정 서브루틴을 양자가 맡아 병목을 줄이는 방식
* 최적화에서 “양자+고전” 혼합 휴리스틱
* 정부 차원의 벤치마킹/검증 프로그램 확대(대규모 투자와 평가 체계) ([The Department of Energy's Energy.gov][3])
#### B구간: 2029–2030년대 초 — 최초의 ‘진짜’ 내결함성 시스템(희소 자원)
로드맵이 모이는 지점이 **2029 전후**입니다.
* IBM: **2029년 Starling**로 **200 논리 큐비트**, **1억(100M) 게이트** 규모 회로 실행을 제시. ([IBM][2])
* Quantinuum: **2030년까지** 범용 내결함성 시스템 목표를 공표. ([Quantinuum][4])
다만 이때도 “모든 기업이 바로 쓰는 시대”가 아니라, 초기 FT 장비는 **비싸고, 희소하고, 예약이 꽉 찬** 형태가 될 가능성이 높습니다.
#### C구간: 2030년대 중반+ — 논리 큐비트 확장과 응용 영역 확대
QEC 오버헤드가 낮아지고 제조·운영이 산업화되면:
* 더 많은 논리 큐비트,
* 더 깊은 회로,
* 더 다양한 산업 응용(소재, 신약, 특정 형태의 최적화/샘플링 등)
으로 이어질 수 있습니다.
---
### 3) 기술 노선이 여러 개인 이유(그리고 전망에 미치는 영향)
* **초전도(IBM/Google 등)**: 빠른 게이트와 생태계 강점. 대신 대규모 배선/극저온/오류정정 비용이 관건. ([IBM][2])
* **이온 트랩(IonQ/Quantinuum)**: 큐비트 품질이 강점. 다만 확장 공학(패키징/멀티플렉싱)이 난제. IonQ는 2030년을 향해 공격적 로드맵을 제시합니다. ([IonQ][5])
* **포토닉(PsiQuantum 등)**: 반도체 공정과 광통신 기반 확장을 노리지만, FT 스택 구축 난이도가 매우 큼. 대규모 시설/투자가 진행 중입니다. ([Reuters][6])
* **위상학(Microsoft)**: 성공하면 하드웨어 레벨에서 오류를 줄여 QEC 부담을 크게 낮출 잠재력. 대신 검증 난이도가 가장 높은 축. Microsoft는 Majorana 1을 로드맵의 단계로 제시합니다. ([Microsoft Azure][7])
* **어닐링(D-Wave)**: 특정 최적화 스타일 업무에 상용 제공을 강조. 범용 게이트형과 성격이 다르고 “이점”은 문제 정의에 크게 의존. ([dwavequantum.com][8])
---
### 4) 어디에 쓸모가 커지나 / 무엇은 당분간 어렵나
**유망도가 가장 높은 축**
1. **화학·소재 시뮬레이션**: 양자계는 양자계로 모사할 때 근본적 이점이 기대됩니다. 구글도 화학·신약을 대표 응용으로 언급합니다. ([Google Research][1])
2. **보안(암호) 영향: “지금 당장 붕괴”보다 “이행이 시급”**
중요한 건 ‘오늘 밤 해킹’이 아니라 **수년 뒤 복호화될 위험(수집 후 추후 복호화)** 때문에 장기 비밀이 위험해진다는 점입니다.
NIST는 **2024년 8월** PQC 표준(ML-KEM/ML-DSA/SLH-DSA)을 확정했고, “지금 사용해도 된다/해야 한다”는 메시지를 명확히 냅니다. ([NIST][9])
3. **최적화/샘플링(선별적)**: 넓게 다 이긴다기보다 “특정 구조의 문제”에서만 승부가 납니다.
**당분간 어려운 것**
* GPU/CPU를 대체해 주류 AI 학습을 맡는 것
* NP-hard 문제를 만능으로 대규모 해결
* 소비자용 ‘데스크톱 양자컴퓨터’(극저온/제어/오류정정이 본질적으로 무겁습니다)
---
### 5) 거시 전망: 국가 전략 인프라로 굳어지는 중
* UN은 **2025년을 ‘세계 양자과학·기술의 해’**로 선포했습니다. ([UN Docs][10])
* 미국 DOE는 **2025년 11월 4일** 국가 QIS 연구센터 5곳에 **6억 2,500만 달러** 재투자를 발표했습니다. ([The Department of Energy's Energy.gov][3])
* EU는 양자 플래그십을 장기 R&D 이니셔티브로 두고 **EU 예산 €10억** 규모를 명시합니다. ([디지털 전략][11])
이제 핵심 병목은 연구만이 아니라, **제조 수율·패키징·극저온 공급망·제어 전자·전력/데이터센터 운영·인재** 같은 산업 요소입니다.
---
### 6) 불확실성을 견디는 실무 전략(추가 팁·응용)
1. **NISQ 파일럿과 FTQC 비즈니스 케이스를 분리**해서 관리해야 합니다.
2. **PQC 이행은 “계획”이 아니라 “프로젝트”**가 되어야 합니다(암호 인벤토리, 장기 비밀 우선, 하이브리드 전환). ([NIST][9])
3. 로드맵의 주류는 “양자 단독”이 아니라 **양자+HPC 결합**입니다. ([IBM][12])
4. “양자 이점”은 대부분 **특정 워크로드·비용모델·오류예산**에 종속됩니다. 따라서 벤치마크는 반드시 **자사 제약조건**과 연결되어야 합니다.
---
## 日本語
### 1) 2026年初頭の到達点
量子計算は依然としてノイズの大きい段階（NISQ）にありますが、焦点は「デモ」から **誤り訂正（QEC）に基づく工学化**へ移りました。
GoogleはWillowで、表面コードの距離を増やすほど符号化誤りが抑制される（スケールで改善する）結果を報告しています。 ([Google Research][1])
IBMは2029年の大規模フォールトトレラント（Starling）を明確に掲げ、モジュール化と実時間デコーディングを含む枠組みを提示しています。 ([IBM][2])
### 2) 時間軸（現実的な3区分）
* **2026–2028**：HPCと組み合わせた“狭い優位（ハイブリッド優位）”が中心。IBMは2026年末までの優位の可能性を述べています。 ([IBM][2])
* **2029〜2030年代初頭**：初期の本格FT機が出るが希少・高価。IBMは2029年に200論理量子ビット/1億ゲート級を掲げています。 ([IBM][2])
* **2030年代中盤以降**：論理量子ビットの規模拡大と応用領域の広がり（ただし産業化が前提）。
### 3) “用途”の見立て（強い順）
* **化学・材料**：量子系を量子で扱う必然性があり有望。 ([Google Research][1])
* **暗号（移行の緊急性）**：NISTは2024年8月にPQC標準（ML-KEM/ML-DSA/SLH-DSA）を確定。 ([NIST][9])
* **最適化**：万能ではなく、問題構造次第。
### 4) 戦略（不確実性に強い）
PQC移行を本格プロジェクト化し、NISQの実験とFTの事業化を切り分け、ハイブリッド（量子+HPC）前提でワークフローを構築するのが最も堅い方針です。 ([IBM][12])
---
## Español
### 1) Estado real (inicio de 2026)
La disciplina está pasando de “demos ruidosas” a **ingeniería con corrección de errores (QEC)**.
Google reportó mejoras de QEC al aumentar el tamaño del código (Willow). ([Google Research][1])
IBM publicó un marco y una hoja de ruta hacia **tolerancia a fallos a gran escala en 2029** (Starling). ([IBM][2])
### 2) Pronóstico por horizontes
* **2026–2028**: “ventajas híbridas” (cuántica + HPC) en casos estrechos; IBM apunta a ejemplos de ventaja alrededor de 2026. ([IBM][2])
* **2029–principios de los 2030**: primeras máquinas tolerantes a fallos, pero escasas; IBM menciona 200 qubits lógicos y 100M compuertas para 2029. ([IBM][2])
* **2030s+**: ampliación de qubits lógicos y portafolio de aplicaciones si la industrialización reduce el overhead.
### 3) Impacto en seguridad (lo más accionable hoy)
NIST finalizó en **agosto de 2024** los primeros estándares de criptografía post-cuántica (ML-KEM, ML-DSA, SLH-DSA) y señala que deben adoptarse ya. ([NIST][9])
---
## Français
### 1) Situation (début 2026)
Le secteur bascule de la phase NISQ vers une **industrialisation guidée par la correction d’erreurs (QEC)**.
Google a présenté des résultats de QEC où l’encodage s’améliore en grandissant (Willow). ([Google Research][1])
IBM vise un jalon clair de **tolérance aux fautes à grande échelle en 2029** (Starling). ([IBM][2])
### 2) Horizons plausibles
* **2026–2028** : gains “hybrides” (quantique + HPC) sur des problèmes étroits ; IBM évoque des exemples d’avantage autour de 2026. ([IBM][2])
* **2029–début 2030s** : premières machines FT, coûteuses et rares ; IBM annonce 200 qubits logiques et 100M portes pour 2029. ([IBM][2])
* **Milieu 2030s+** : élargissement des cas d’usage si la mise à l’échelle devient manufacturable.
### 3) Le point le plus concret dès maintenant : la transition crypto
NIST a finalisé en **août 2024** les premiers standards post-quantiques (ML-KEM, ML-DSA, SLH-DSA) et recommande de les déployer. ([NIST][9])
---
* [Reuters](https://www.reuters.com/business/psiquantum-breaks-ground-chicago-quantum-site-after-1-billion-funding-2025-09-30/?utm_source=chatgpt.com)
* [CT Insider](https://www.ctinsider.com/business/article/quantum-technology-yale-uconn-quantumct-lamont-21197360.php?utm_source=chatgpt.com)
* [statesman.com](https://www.statesman.com/business/technology/article/ut-semiconductor-chips-grant-greg-abbott-21235615.php?utm_source=chatgpt.com)
* [barrons.com](https://www.barrons.com/articles/d-wave-quantum-computing-stock-d8596186?utm_source=chatgpt.com)
[1]: https://research.google/blog/making-quantum-error-correction-work/ "Making quantum error correction work"
[2]: https://www.ibm.com/quantum/blog/large-scale-ftqc "IBM lays out clear path to fault-tolerant quantum computing | IBM Quantum Computing Blog"
[3]: https://www.energy.gov/articles/energy-department-announces-625-million-advance-next-phase-national-quantum-information?utm_source=chatgpt.com "Energy Department Announces $625 Million to Advance ..."
[4]: https://www.quantinuum.com/press-releases/quantinuum-unveils-accelerated-roadmap-to-achieve-universal-fault-tolerant-quantum-computing-by-2030?utm_source=chatgpt.com "Quantinuum Unveils Accelerated Roadmap to Achieve ..."
[5]: https://www.ionq.com/roadmap?utm_source=chatgpt.com "Roadmap"
[6]: https://www.reuters.com/business/psiquantum-breaks-ground-chicago-quantum-site-after-1-billion-funding-2025-09-30/?utm_source=chatgpt.com "PsiQuantum breaks ground on Chicago quantum site after $1 billion funding"
[7]: https://azure.microsoft.com/en-us/blog/quantum/2025/02/19/microsoft-unveils-majorana-1-the-worlds-first-quantum-processor-powered-by-topological-qubits/?utm_source=chatgpt.com "Microsoft unveils Majorana 1, the world's first quantum ..."
[8]: https://www.dwavequantum.com/solutions-and-products/systems/?utm_source=chatgpt.com "The Advantage2™ Quantum Computer"
[9]: https://www.nist.gov/news-events/news/2024/08/nist-releases-first-3-finalized-post-quantum-encryption-standards "NIST Releases First 3 Finalized Post-Quantum Encryption Standards | NIST"
[10]: https://docs.un.org/en/A/RES/78/287?utm_source=chatgpt.com "A/RES/78/287 - General Assembly - the United Nations"
[11]: https://digital-strategy.ec.europa.eu/en/policies/quantum-technologies-flagship?utm_source=chatgpt.com "Quantum Technologies Flagship"
[12]: https://www.ibm.com/roadmaps/quantum/ "www.ibm.com"