Networking in the Age of AI
Ashkan Seyedi, NVIDIA
As AI systems scale to trillions of parameters and exascale compute clusters, the role of networking has never been more critical. Modern AI workloads demand extreme bandwidth, ultra‑low latency, and highly reliable interconnects to efficiently move vast volumes of data across accelerated compute fabrics. This talk will explore how next‑generation networking architectures—driven by advances in optical interconnects, co‑packaged optics, and energy‑efficient link technologies—are reshaping the performance and sustainability of AI infrastructure. I will highlight the importance of low‑power, high‑reliability optical communication in enabling scalable, resilient, and environmentally responsible AI systems, and discuss how these technologies form the backbone of tomorrow’s AI factories.
Redefining Future Coherent Transceivers via Subsystem, Device, and DSP Innovations
Xi (Vivian) Chen, Nokia Bell Labs
This talk aims to discuss how innovations at the system and subsystem levels could improve the design of coherent optical transceivers, addressing the growing demand for interface rates, form factors, and power consumption.
From Lab to Live Networks: Supercharging UK Telecoms R&I with JOINER”
Dimitra Simeonidou, University of Bristol
Future telecoms innovation depends less on isolated prototypes and more on large-scale, realistic experimentation that can validate performance, security, resilience and deployability across diverse technology environments: from optical and satellite to wireless, compute and cloud. This keynote introduces JOINER, the UK-scale federated experimentation platform designed to accelerate telecoms research and innovation by connecting multiple sites, facilities and partners into a single, repeatable test environment.
I will explain how JOINER enables end-to-end experimentation across multi-technology open and disaggregated networks, AI-enabled network operations, and advanced security capabilities, bridging the gap between lab results and deployable solutions. The talk will highlight how JOINER supports faster onboarding and trial cycles, evidence-based assurance, and collaboration across universities, industry, SMEs and government, turning research outputs into pilots and real-world adoption.
Using concrete examples, I will show how experimentation at scale de-risks technology choices, shortens time-to-impact, and strengthens the UK’s position in international telecoms alliances. I will close with what’s needed next: shared metrics and practices, sustainable federated infrastructure, and an open innovation model that makes large-scale experimentation the default pathway from research to deployment.
The Networked Quantum Data Machine: Scaling Secure Services through Photonic MBQC
Navigating the Scalability-Privacy-Integrity-Utility Challenge
Elham Kashefi, NQCC/University of Edinburgh/CNRS
As we transition toward distributed quantum computing as a service in 2026, the primary challenge lies in scaling secure services without compromising privacy, integrity, or utility. This keynote explores why Measurement-Based Quantum Computing serves as the foundational architecture for this shift, leveraging measurement-driven adaptivity to bridge the gap between high-level verification protocols and the physical constraints of modern optical technologies. Drawing on our latest 2025-2026 breakthroughs, I will detail the roadmap for the Networked Quantum Data Machine, specifically demonstrating how Adaptive State Injection grants linear photonic circuits the non-linear expressive power required for Quantum Machine Learning to achieve a proven polynomial advantage over classical models. By integrating software-hardware co-design, from information-theoretic verification protocols on Qline architectures to the optimization of Isolated Gate Zones on-chip, we establish a robust framework for realizing the full potential of secure, verifiable, and networked quantum computing.