AI continues to come at us from all directions. Imagine a world where billions of people, AI agents, and operators are in constant communication, exchanging data and solving previously unthinkable problems at lightning speed.

The change is not limited to the speed of AI evolution; it's also about the magnitude of complexity it will introduce. So, what do organizations need to do differently to prepare for this? AI transformation demands high-speed interconnects both within and between data centers-which only fiber optics can provide. Effectively navigating AI requires a fundamental architectural shift, and Cisco is here for it.

Maximizing reliability and performance for AI data center connections

With AI investments expected to surge to$5.2 trillion by 2030 according to McKinsey Quarterly, the demands on networks have never been greater. Data center traffic is exploding, with back-end traffic growing 10x every two years and the rapid adoption of 800G and 1.6T speeds as reported by the Dell'Oro Group. AI clusters are scaling at an unprecedented pace, with the largest models now exceeding 100,000 GPU nodes-over 20x larger than just a few years ago according to Epoch AI.

Amid such a dramatic shift in data center infrastructure, reliability isn't just important, it's essential. An analysis from Meta found that a single slow GPU link or failed network connection can reduce cluster performance by up to 40%, leaving expensive GPUs idle and investments stranded. Unlike traditional IP networking, AI workloads require the highest levels of link integrity to ensure continuous training and inference. This puts immense pressure on high-speed optics, which must operate under the intense conditions found in today's AI data centers.

Cisco is uniquely positioned to meet these challenges. We have what we like to call a "platform advantage" in the full stack of technology-ASICs, optics, platforms, software, security, and visibility. This allows us to optimize the design and integration of every element to ensure performance and reliability for AI environments.

Cisco's strategy for AI infrastructure relies on three core strengths:

Cisco silicon photonics: foundation, performance, and reliability

• At the heart of our innovation is Cisco silicon photonics, delivering higher performance, better power efficiency, and increased reliability. Cisco has over two decades of high-volume silicon photonics manufacturing and deployment experience.

Industry's most robust stress testing for demanding AI environments

• Leveraging our systems and optics design expertise, we stress test our optics for AI networking parameters. Our testing methodology encompasses optical, electrical, and mechanical design verification; regulatory and standards compliance; as well as Cisco and third-party host platform and software compatibility testing. Rigorous qualification is performed at the transceiver module level and at the platform level of fully loaded switches and routers.

Architectural innovations to reduce power consumption

• At Cisco Live this past June, we demonstrated a 51.2T switch populated with 64 x 800G linear pluggable optics. By shifting power-consuming DSP functionality from pluggable optics to the Cisco Silicon One ASIC, we demonstrated a 30% reduction in system power compared to traditional retimed optics.
• We are also committed to emerging technologies such as co-packaged optics (CPO) that can further reduce overall SerDes power consumption required to generate the optical channel due to ultra-short electrical traces. Following capability demonstrations in 2023, our approach to this emerging technology/solution includes leveraging our global supply chain for diversity in components from multiple foundries, optical engines, and geographies.

In addition, Cisco's globally diversified supply chain leverages multiple sources for silicon photonics wafer fabs, DSPs, lasers, and manufacturing. This ensures timely access to critical components, enhancing supply chain resilience and enabling rapid deployments for the most demanding AI data center infrastructure needs.

Connecting AI data centers with coherent transport

As AI data center infrastructure continues to grow, it has created demand for connecting data centers using coherent transport. AI's impact on coherent transport demand is undisputable, and Cisco can deliver the bandwidth required for this new generation of computing. Large language model (LLM) version updates and higher performing GPUs have been introduced, terminology such as neocloud provider and AI factories have become increasingly mainstream, and new entities are being created worldwide to house the infrastructure to create massive AI training facilities.

In addition, multiple service providers that have traditionally supported wholesale bandwidth and fiber connectivity services have publicly announced deals to support AI connectivity growth. These connections would specifically address connectivity between data center sites-with the AI connectivity opportunities not limited to hyperscaler sites, but also including enterprise customer connectivity.

The ramp-up of AI networks and infrastructure is playing a part in current 400G coherent pluggable module adoption, as well as expected 800G coherent module adoption. Cignal AI recently stated that growth in recent quarters of coherent pluggable 400G shipments is being driven by data center interconnects (DCIs) for AI data centers, with AI software being designed to support multiple physical locations. As hyperscalers and service providers build out to support AI, Cisco coherent modules are already being deployed in these DCI builds, and customers are projecting increasing bandwidth requirements to support these architectures.

Cisco Routed Optical Networking continues to expand and becomes even more critical in the AI era. According to Heavy Reading, a significant portion (58%) of communication service providers plan to deploy 800G coherent pluggables between 2026 and 2028. New Cisco coherent modules, including 100ZR, 400G ultra long-haul (ULH), and 800G ZR/ZR+, are enabling 100% coverage of network applications from shorter reach campus and access applications to long-haul and even subsea, as evidenced by groundbreaking trials.

AARNet (Australia's Academic and Research Network) achieved a remarkable milestone with the successful completion of a field trial showcasing the capabilities of cutting-edge 400G ULH pluggable coherent optics technology. Conducted over an incredible distance of 4,600 km on the Indigo Central subsea cable system, which connects Sydney and Perth, this achievement heralds a new era in long-haul optical networking.

G