26-9 ECOC 2026 개최
작성자의 핵심 판단
작성자는 전기 인터커넥트만으로 감당하기 어려운 AI 컴퓨트 확장 속에서 광연결이 네트워크 인프라를 넘어 컴퓨팅 인프라로 들어가는 흐름이라고 본다.
근거 보기 4
- 원문 구간 1
[원문 유형] CEO 분석용 텔레그램 채널별 뷰 원문 [채널] 황금똥밭 연간 3기방 [묶음 주기] 12시간 [통합 구간] 2026-09-22T03:00:00+00:00 ~ 2026-09-22T15:00:00+00:00 [작성 기준] 아래 글은 모두 같은 유료 채널 작성자가 시간 순서대로 이어 낸 관점이다. 여러 글을 단순 나열하지 말고, 무엇을 사실 근거로 삼았는지, 작성자가 어떻게 해석했는지, 전망·확신·조건이 구간 안에서 어떻게 이어지거나 달라졌는지를 구분해 하나의 문서로 정리한다. 외부 링크와 첨부는 작성자 판단의 근거 원문이며, 외부 원문의 주장과 작성자 자신의 뷰를 섞지 않는다. 시간흐름순 정리에서는 게시 순서와 판단 강도, 유보 조건, 인상적인 표현을 보존한다. ===== 통합 원문 1 / 13 ===== [채널] 황금똥밭 연간 3기방 [게시일시] 2026.09.22 12:12:40 [텔레그램 게시물] https://t.me/c/3931993375/1463 [원문 수집 기준] 외부 링크가 있으면 링크를 따라 확보한 실제 본문·첨부 파일이 주 분석 원문이다. 텔레그램 게시물은 발견 경로와 보조 설명으로 보존하며 외부 원문과 구분해 반영한다. [현재 텔레그램 게시물 본문] 제목: ECOC 2026 Plenary Takeaways: How Ciena, imec, Huawei, and PsiQuantum Are Redefining Optical Infrastructure for the AI Era 보낸이: SEMIVISION 날짜: 2026년 9월 21일 ECOC 2026의 4개 기조연설이 하나의 기술 로드맵으로 모였다는 분석이에요. Ciena (Peter Winzer): GPU/XPU 클러스터 확대로 광 연결이 컴퓨트에 더 가까워지는 흐름을 AI 클러스터 아키텍처 관점에서 설명. Scale-up, scale-out, scale-across 구조에서 광의 역할 강조.
- 원문 구간 2
SemiVision believes these four presentations become much more meaningful when viewed together rather than as separate discussions about optical transceivers, silicon photonics, or CPO. From system architecture and semiconductor technology to optical packaging and computing architecture, they collectively address the same fundamental question: As AI compute scaling begins to exceed what electrical interconnects can efficiently support, how will photonics evolve from network infrastructure into computing infrastructure? Ciena — AI Cluster Communications Peter Winzer imec — Integrated Photonics at the Heart of a Scalable AI Revolution Patrick Vandenameele Huawei — Scaling AI: From Optical Innovation to NPO and CPO Solutions Jiangwei Man PsiQuantum — Photonics for Quantum Computing Mark Thompson ECOC 2026 Day 1: In the AI Era, Optical Communications Is Shifting from a Race for Speed to System-Level Integration Sep 21 Read full story
- 원문 구간 3
From system architecture and semiconductor technology to optical packaging and computing architecture, they collectively address the same fundamental question: As AI compute scaling begins to exceed what electrical interconnects can efficiently support, how will photonics evolve from network infrastructure into computing infrastructure? Ciena — AI Cluster Communications Peter Winzer imec — Integrated Photonics at the Heart of a Scalable AI Revolution Patrick Vandenameele Huawei — Scaling AI: From Optical Innovation to NPO and CPO Solutions Jiangwei Man PsiQuantum — Photonics for Quantum Computing Mark Thompson ECOC 2026 Day 1: In the AI Era, Optical Communications Is Shifting from a Race for Speed to System-Level Integration Sep 21 Read full story Continue reading this post for free, courtesy of SEMIVISION. Claim my free post Or purchase a paid subscription. Previous [외부 원문 3] [URL] https://tspasemiconductor.substack.com/p/ecoc-2026-plenary-takeaways-how-ciena
- 원문 구간 4
As AI compute scaling begins to exceed what electrical interconnects can efficiently support, how will photonics evolve from network infrastructure into computing infrastructure? Ciena — AI Cluster Communications Peter Winzer imec — Integrated Photonics at the Heart of a Scalable AI Revolution Patrick Vandenameele Huawei — Scaling AI: From Optical Innovation to NPO and CPO Solutions Jiangwei Man PsiQuantum — Photonics for Quantum Computing Mark Thompson ECOC 2026 Day 1: In the AI Era, Optical Communications Is Shifting from a Race for Speed to System-Level Integration Sep 21 Read full story Continue reading this post for free, courtesy of SEMIVISION. Claim my free post Or purchase a paid subscription. Previous ===== 통합 원문 3 / 13 ===== [채널] 황금똥밭 연간 3기방 [게시일시] 2026.09.22 12:20:47 [텔레그램 게시물] https://t.me/c/3931993375/1465 [원문 수집 기준] 외부 링크가 있으면 링크를 따라 확보한 실제 본문·첨부 파일이 주 분석 원문이다.
이 자료에서 다룬 사실
ECOC 2026 기조연설은 Ciena의 AI 클러스터 통신, imec의 실리콘 포토닉스·3D 집적·첨단 패키징, Huawei의 NPO·CPO·optical I/O 선택, PsiQuantum의 양자컴퓨터용 포토닉스를 각각 다뤘다.
근거 보기 4
- 원문 구간 1
[원문 유형] CEO 분석용 텔레그램 채널별 뷰 원문 [채널] 황금똥밭 연간 3기방 [묶음 주기] 12시간 [통합 구간] 2026-09-22T03:00:00+00:00 ~ 2026-09-22T15:00:00+00:00 [작성 기준] 아래 글은 모두 같은 유료 채널 작성자가 시간 순서대로 이어 낸 관점이다. 여러 글을 단순 나열하지 말고, 무엇을 사실 근거로 삼았는지, 작성자가 어떻게 해석했는지, 전망·확신·조건이 구간 안에서 어떻게 이어지거나 달라졌는지를 구분해 하나의 문서로 정리한다. 외부 링크와 첨부는 작성자 판단의 근거 원문이며, 외부 원문의 주장과 작성자 자신의 뷰를 섞지 않는다. 시간흐름순 정리에서는 게시 순서와 판단 강도, 유보 조건, 인상적인 표현을 보존한다. ===== 통합 원문 1 / 13 ===== [채널] 황금똥밭 연간 3기방 [게시일시] 2026.09.22 12:12:40 [텔레그램 게시물] https://t.me/c/3931993375/1463 [원문 수집 기준] 외부 링크가 있으면 링크를 따라 확보한 실제 본문·첨부 파일이 주 분석 원문이다. 텔레그램 게시물은 발견 경로와 보조 설명으로 보존하며 외부 원문과 구분해 반영한다. [현재 텔레그램 게시물 본문] 제목: ECOC 2026 Plenary Takeaways: How Ciena, imec, Huawei, and PsiQuantum Are Redefining Optical Infrastructure for the AI Era 보낸이: SEMIVISION 날짜: 2026년 9월 21일 ECOC 2026의 4개 기조연설이 하나의 기술 로드맵으로 모였다는 분석이에요. Ciena (Peter Winzer): GPU/XPU 클러스터 확대로 광 연결이 컴퓨트에 더 가까워지는 흐름을 AI 클러스터 아키텍처 관점에서 설명. Scale-up, scale-out, scale-across 구조에서 광의 역할 강조.
- 원문 구간 2
f Ciena approached the problem from the perspective of AI cluster architecture, examining how the continued expansion of GPU and XPU clusters is forcing optical connectivity closer to compute through scale-up, scale-out, and scale-across architectures. Patrick Vandenameele, CEO of imec, addressed the same challenge from a semiconductor technology perspective, arguing that future AI scaling can no longer rely on CMOS alone. Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer.
- 원문 구간 3
Patrick Vandenameele, CEO of imec, addressed the same challenge from a semiconductor technology perspective, arguing that future AI scaling can no longer rely on CMOS alone. Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer. SemiVision believes these four presentations become much more meaningful when viewed together rather than as separate discussions about optical transceivers, silicon photonics, or CPO.
- 원문 구간 4
Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer. SemiVision believes these four presentations become much more meaningful when viewed together rather than as separate discussions about optical transceivers, silicon photonics, or CPO. From system architecture and semiconductor technology to optical packaging and computing architecture, they collectively address the same fundamental question:
그렇게 판단한 이유
GPU·XPU 클러스터 확대로 광 연결이 컴퓨트에 가까워지고, 224G-per-lane 이후 전기 인터커넥트의 부담이 커지는 점을 하나의 시스템 전환으로 연결한다.
근거 보기 4
- 원문 구간 1
imec (Patrick Vandenameele): 반도체 기술 관점에서, 앞으로 AI 스케일링은 CMOS만으로는 안 되며 실리콘 포토닉스, 3D 집적, 첨단 패키징이 하나의 플랫폼으로 진화해야 한다고 주장. Huawei (Jiangwei Man): 224G-per-lane 시대로 가면서 NPO, CPO, 궁극적으로 optical I/O 중 어떤 선택을 할지가 시스템 아키텍처의 핵심이라는 내용. PsiQuantum (Mark Thompson): 빛을 프로세서 간 정보 이동이 아니라 대규모 양자컴퓨터의 물리적 기반으로 쓰는 방향을 제시. 전기 인터커넥트로 감당하기 어려운 AI 컴퓨트 스케일링 시대에, 포토닉스가 네트워크 인프라에서 컴퓨팅 인프라로 진화하는 과정이라는 거 광연결은 시대의 흐름 + 양자 역시 마찬가지. 광자 방식을 택하면 광연결은 필수가 맞고, 다른 방식이라도 분산·대규모화를 가면 광 네트워킹은 사실상 피할 수 없는 방향 ===== 통합 원문 2 / 13 ===== [채널] 황금똥밭 연간 3기방 [게시일시] 2026.09.22 12:13:32 [텔레그램 게시물] https://t.me/c/3931993375/1464 [원문 수집 기준] 외부 링크가 있으면 링크를 따라 확보한 실제 본문·첨부 파일이 주 분석 원문이다. 텔레그램 게시물은 발견 경로와 보조 설명으로 보존하며 외부 원문과 구분해 반영한다. [현재 텔레그램 게시물 본문] [첨부 미디어 게시물] [https://tspasemiconductor.substack.com/p/ecoc-2026-plenary-takeaways-how-ciena?utm_campaign=post-expanded-share&utm_medi] https://tspasemiconductor.substack.com/p/ecoc-2026-plenary-takeaways-how-ciena?utm_campaign=post-expanded-share&utm_medium=web
- 원문 구간 2
f Ciena approached the problem from the perspective of AI cluster architecture, examining how the continued expansion of GPU and XPU clusters is forcing optical connectivity closer to compute through scale-up, scale-out, and scale-across architectures. Patrick Vandenameele, CEO of imec, addressed the same challenge from a semiconductor technology perspective, arguing that future AI scaling can no longer rely on CMOS alone. Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer.
- 원문 구간 3
Patrick Vandenameele, CEO of imec, addressed the same challenge from a semiconductor technology perspective, arguing that future AI scaling can no longer rely on CMOS alone. Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer. SemiVision believes these four presentations become much more meaningful when viewed together rather than as separate discussions about optical transceivers, silicon photonics, or CPO.
- 원문 구간 4
Silicon photonics, 3D integration, advanced packaging, and electronics will increasingly need to evolve as part of a unified technology platform. Jiangwei Man of Huawei brought the discussion closer to practical system architecture, focusing on the choices among NPO, CPO, and ultimately optical I/O as electrical interconnects move into the 224G-per-lane generation and beyond. Finally, PsiQuantum co-founder and CTO Mark Thompson extended integrated photonics into an even more fundamental computing application: light is no longer used only to move information between processors, but can become part of the physical foundation of a large-scale quantum computer. SemiVision believes these four presentations become much more meaningful when viewed together rather than as separate discussions about optical transceivers, silicon photonics, or CPO. From system architecture and semiconductor technology to optical packaging and computing architecture, they collectively address the same fundamental question: