Goldman Sachs report: Optical Networking is the next mega trend in AI infrastructure
Goldman Sachs analysts forecast a $154billion opportunity in optical networking driven by skyrocketing capacity demands from hyperscale cloud and AI workloads. Carriers and vendors are integrating 10GbE edge networking and AI-RAN (Artificial Intelligence Radio Access Network) trials on live 5G networks.
Goldman argues that AI infrastructure is creating a networking bottleneck phase, where optical interconnects become essential to connect more chips, keep latency low, and let AI clusters scale efficiently. The total optical networking market forecast 9x increase to $154 billion is due to both scale-up and scale-out AI data center architectures grow.
AI compute gains are no longer just about faster GPU and HBM chips; they depend on moving data fast enough between chips, racks, and super-nodes. Goldman Sachs emphasizes that networking now “unlocks computing capability” by enabling seamless exchange across multiple AI chips, which is exactly where copper-based links start to fall short. That makes fiber-optic connectivity, pluggable optics, and co-packaged optics central to the next phase of AI build-out. The report splits opportunity across scale-up and scale-out networking, plus component categories such as copper cables, pluggable optical modules, CPO, and PCB midplanes. External coverage of the report says Goldman Sachs sees scale-up as the larger pool, about $106 billion or 69% of the $154 billion TAM, while CPO could represent about $91 billion or 59% of the total, assuming 29% penetration in scale-out networking. In practical terms, the report is signaling that the highest-value optical opportunity sits inside tightly coupled AI systems, not just in long-haul or metro transport.
………………………………………………………………………………………………………………………………………………………………………………………….
Goldman projects the following:
- Dollar content increase by 16x / 45x in Scale Out / Scale Up per computing unit from GB300 NVL72 (per computing unit means 72 GPUs per rack to reach NVL72) to Rubin Ultra NVL576 (per computing unit means 72 GPUs per rack, and 8 racks together to reach NVL576), with opportunities across pluggable optical modules, optical engines in CPO, copper cables, and PCB midplanes.
- A 13x larger addressable market for optical modules / optical engines expanding from scale out (e.g. GB300 NVL72) to scale up (e.g. Nvidia Rubin Ultra [1.] NVL576 level 2 scale up via CPO) per computing unit. n
- A 10x larger value market for pluggable optical modules in scale out per computing unit from GB300 NVL72 to Rubin Ultra NVL576, even with a 29% CPO penetration rate. The numbers of pluggable optical module (1.6T equivalent) per computing unit would increase from 216 units in GB300 NVL72 to 2.5k units in Rubin Ultra NVL576.
Note 1. Nvidia Rubin Ultra is a flagship, next-generation AI and high-performance computing (HPC) processor succeeding the standard Rubin architecture. Scheduled to debut in late 2027, it utilizes massive multi-die chiplet designs and unprecedented memory configurations to power the next wave of generative and agentic AI.
………………………………………………………………………………………………………………………………………………………………………….
Market Forecasts:
The investment bank expects the aggregate dollar content per computing unit across scale up and scale out to increase by 29x from US$315k in GB300 NVL72 to US$9.4bn in Rubin Ultra NVL576, and assuming the numbers of racks through the full product cycle are 48k racks for GB300 NVL72, and 16.5k computing units for Rubin Ultra NVL576, the aggregate value TAM across scale up and scale out would increase by 9x from US$15bn in GB300 NVL72 (mainly in 2026) to US$154bn in Rubin Ultra NVL576 (mainly in 2028).
Among the US$154bn value TAM, 69% goes to scale up, or US$106bn, and CPO contributes US$91bn, or 59% of the US$154bn value TAM, assuming CPO at 29% penetration rate in scale out.
For network architects, the important takeaway is that AI clusters are becoming optics-heavy at more layers of the network stack, not just at the edge of the rack. The likely winners are suppliers that can reduce power, improve density, and simplify packaging for very high-bandwidth links, especially around CPO and advanced pluggables. This is less a story about traditional telecom optics and more about datacenter interconnects optimized for GPU fabrics and AI training/inference throughput.
The most consistently cited “top beneficiaries” are Coherent, Lumentum, and Fabrinet. These companies sit close to the core optical component modules and manufacturing layers that scale with higher AI interconnect demand. That makes them the most straightforward proxies for the forecasted optics expansion. The report’s thesis favors companies with strong exposure to high-end optical transport, coherent optics, and data-center interconnect rather than the broader optical networking/PON equipment companies like Ciena, Nokia/Infinera, Cisco/Acacia, ADVA, or Calix.
Conclusions:
Strategically, Goldman Sachs maintains that optical networking is no longer a niche enabling layer; it is becoming a core enabler of AI capex scaling. That shifts investor attention toward optical component vendors, silicon photonics, transceiver suppliers, and adjacent packaging ecosystems. The report’s core message is simple: as AI clusters grow, the network fabric becomes a first-order constraint, and optics are the most likely answer.
References:
2026 Fiber Connect Keynote: “The Future of Fiber Optics: AI and the Quantum”
How will fiber and equipment vendors meet the increased demand for fiber optics in 2026 due to AI data center buildouts?
Big Fiber’s $250M financing deal to buildout dark fiber routes for AI Data Center expansion
Analysis: Fiber Broadband Association (FBA) whitepaper: Upgrading MSO Networks to Fiber to the Home (FTTH): A Technical Perspective
Fiber Optic Boost: Corning and Meta in multiyear $6 billion deal to accelerate U.S data center buildout
Fiber Optic Networks & Subsea Cable Systems as the foundation for AI and Cloud services
AI infrastructure investments drive demand for Ciena’s products including 800G coherent optics
DriveNets and Ciena Complete Joint Testing of 400G ZR/ZR+ optics for Network Cloud Platform
