Goldman Sachs and Nokia are both pointing to photonics—using light instead of electricity to move data—as a critical upgrade path for artificial intelligence data centers. The emphasis comes as AI clusters expand and investors look for solutions to the emerging “connectivity” bottleneck, where electrical links struggle with speed, heat and energy efficiency. Nokia says it is ramping silicon-photonics capabilities, aligning with Goldman’s view that optical networking is set to become the next major spending cycle around AI infrastructure.
Key takeaways
-
The articles highlight a potential shift in AI infrastructure spending toward optical networking and photonic components, with Nokia positioning for growth in silicon photonics.
-
Catalyst: Nokia is expanding advanced test and packaging operations for photonic chips, while Goldman Sachs forecasts optical networking demand tied to AI scaling rapidly.
-
Key implication: As AI training and scaling become more connectivity-limited, optical links may be prioritized over incremental improvements to electrical interconnects.
-
Nokia’s financial disclosures suggest momentum in its AI and cloud exposure and optical networks unit, supporting its strategic direction.
What photonics changes for AI data centers
Photonics refers to generating, controlling and detecting light to transmit information. In the context of silicon photonics, components such as lasers, modulators and detectors can be integrated directly onto chips using semiconductor manufacturing processes. Compared with conventional electrical wiring, optical interconnects are designed to support higher bandwidth, lower latency and reduced power consumption.
Goldman Sachs’ framing is that the next constraint in AI development may be less about raw compute and more about how efficiently data moves between chips and racks. As GPU clusters grow, electrical signals can require more energy and generate more heat to deliver the needed throughput. Optical links are positioned as an alternative that can scale with dense data center layouts while improving energy efficiency.
Goldman Sachs sees optical networking as the next AI infrastructure bottleneck
According to the report, Goldman Sachs characterizes optical networking as the next mega-trend in AI infrastructure. The investment bank projects rapid expansion in the total addressable market for optical networking tied to AI, with forecasts spanning from roughly $15 billion in 2026 to $154 billion by 2028.
Goldman segments the opportunity into “scale-up networking” and related optical subcategories. Scale-up networking—high-bandwidth connections within racks—is presented as the dominant slice of the total forecast. Within that area, co-packaged optics (CPO), which integrates optical engines with processors for efficiency gains, are estimated to be a substantial standalone opportunity at scale.
For investors, the practical takeaway is that AI infrastructure upgrades may not end at faster GPUs. Data center operators and suppliers may need optical switching and interconnect solutions to sustain scaling economics as training workloads and inference demands intensify.
Nokia’s photonics push: from strategy to manufacturing expansion
Nokia is attempting to translate the photonics thesis into capacity and execution. The company recently announced that it is expanding advanced test and packaging operations in Allentown, Pennsylvania, focused on photonic chips used in optical modules serving both AI and telecom networks.
Separately, Nokia’s first-quarter 2026 results were cited as showing progress across AI infrastructure-related businesses. According to the company’s disclosures referenced in the article, Nokia’s AI and Cloud segment increased 49% year over year and represents 8% of total sales. The Optical Networks segment grew 20%, supported by AI & Cloud design wins for both pluggables and line systems. Nokia also booked 1 billion euros of new AI infrastructure orders in the first quarter alone.
While those figures point to demand traction, investors will likely focus on whether design wins convert into sustained revenue and whether Nokia can secure reliable supply-chain capacity for photonic components as AI data center build-outs accelerate.
Market reaction and what to watch
There is no market pricing information in the source material provided, so the focus remains on the strategic direction and reported execution. Still, the narrative aligns with how investors have approached AI infrastructure: the winners are increasingly expected to be those supplying the plumbing that enables hyperscale compute growth—especially at the connectivity layer.
Looking ahead, investors in Nokia and the broader optical networking supply chain may watch for updates on manufacturing scale-up in photonic packaging and test operations, additional AI & Cloud design win announcements, and any signs that optical interconnect demand is broadening beyond early deployments into larger, repeatable procurement cycles. Upcoming corporate earnings and further guidance on order conversion, margins, and capacity utilization will be key in assessing whether the photonics shift becomes a durable revenue driver.
