Key Takeaways
- Samsung’s networks revenue plunged in Q1 2026 to $407.5 million, down sharply from $543 million in Q1 2025 and $734 million in Q1 2023.
- The broader telecom market is weak; Ericsson’s mobile‑network sales fell 8 % and Nokia’s dropped 3 % in the same quarter.
- Samsung is shifting focus from legacy mobile networks to silicon photonics and co‑packaged optics (CPO), aiming to supply these components to hyperscale AI and data‑center customers.
- The company’s Foundry division will begin mass‑production of optical modules for a major player in the second half of 2026, leveraging its semiconductor expertise rather than selling its own optical gear.
- Samsung’s memory business is also surging, with HBM4 shipments already underway and HBM4E samples slated for Q2; HBM sales are projected to grow >3× year‑on‑year in 2026.
- While the networks division remains a small, poorly highlighted segment, Samsung hints at using its leadership in vRAN, O‑RAN and AI‑RAN to capture new markets later in the year.
- Overall, Samsung is positioning itself as an AI‑infrastructure powerhouse by combining photonics, advanced memory, and heterogeneous compute capabilities.
Declining Networks Revenue Signals a Turning Point
Samsung’s networks business took a noticeable hit in the first quarter of 2026, reporting revenue of $407.5 million (600 billion KRW). This figure marks a substantial drop from roughly $543 million (800 billion KRW) in Q1 2025 and an even steeper decline compared with $734 million (1.08 trillion KRW) recorded in Q1 2023. The downturn reflects a broader slowdown in mobile‑network spending as the industry wrestles with the lull between 5G roll‑outs and the nascent 6G ecosystem. As one industry observer noted, “Samsung’s networks revenue fell sharply year on year in Q1 2026,” underscoring the urgency for the conglomerate to recalibrate its growth engines.
Telecom Market Headwinds Hit Rivals Too
Samsung is not navigating these challenges alone. Ericsson reported an 8 % decline in mobile‑network sales for Q1 2026, while Nokia’s comparable figure fell 3 %. Ericsson’s overall revenue also slipped 10 %, though Nokia managed a modest 2 % increase, largely propelled by its Optical Networks division and a surge in sales to AI and cloud customers. These numbers illustrate a sector‑wide pressure point: traditional telecom equipment vendors are seeing diminishing returns from legacy mobile infrastructure, prompting them to hunt for higher‑growth adjacencies such as optical interconnects and AI‑centric solutions.
Samsung’s Pivot Toward Optical Technologies
Recognizing the shift, Samsung has begun to “get the memo that optical is the way to go as AI drives demand among hyperscale customers.” In March, the company announced that its Foundry division would enter the silicon photonics and co‑packaged optics (CPO) arena. During the Q1 earnings call, Foundry EVP Suk Chae Kang outlined the plan, stating:
“We are actively pursuing structural transformation by diversifying our application portfolio beyond mobile into AI HPC, automotive, and aerospace sectors… For data centers, we are seeing growing demand for high-bandwidth and low-power data transmission, which has led to a rapid rise in demand for silicon photonics. We are developing not only silicon photonics components, but also technology for the CPO or co‑packaged optics business.”
Kang’s comments reveal a clear strategic pivot: Samsung intends to manufacture photonics and CPO solutions for third‑party customers rather than selling its own branded optical products. This approach mirrors the contract‑manufacturing model that has served its semiconductor foundry well, allowing Samsung to capitalize on the exploding need for high‑speed, low‑latency interconnects inside AI servers and data‑center switches.
Silicon Photonics and CPO: A Different Play from Nokia
While Nokia is boosting its own Optical Networks division to capture AI‑driven demand, Samsung’s route diverges. The South Korean giant will act as a supplier of silicon photonics dies, hybrid integration processes, and CPO assembly services to optical communications module makers. By focusing on the fabrication and packaging layers, Samsung leverages its leadership in advanced process nodes (e.g., 3 nm GAA) and its extensive experience with heterogeneous integration. This model reduces the need for Samsung to build a full‑stack optical product line while still enabling it to reap the margin benefits of supplying critical building blocks to hyperscalers such as Google, Microsoft, and Amazon.
Memory Business Provides an Additional AI Tailwind
Samsung’s pivot is not limited to optics; its memory segment is also experiencing a boom driven by AI workloads. The company began mass‑producing HBM4 in Q1 2026 and plans to ship HBM4E samples in Q2. Memory EVP Jaejune Kim highlighted the outlook on the earnings call:
“We expect HBM sales to increase substantially by more than threefold year on year in 2026.”
The surge reflects the growing appetite for high‑bandwidth memory (HBM) as AI training models swell in size and complexity. With memory bandwidth often becoming the bottleneck in AI accelerators, Samsung’s early leadership in HBM positions it to capture a sizable share of the AI‑hardware market, complementing its photonics ambitions by delivering both the compute‑side interconnects and the memory subsystem that feed those accelerators.
Broader Vision: Becoming an AI Infrastructure Powerhouse
Taken together, Samsung’s moves signal a holistic ambition: to become a one‑stop shop for AI‑centric infrastructure. By combining advanced logic (via its foundry), cutting‑edge memory (HBM), and emerging optical interconnects (silicon photonics/CPO), the company aims to address the three primary constraints in modern AI systems—compute density, memory bandwidth, and data‑movement efficiency. The phrase from the earnings release that Samsung will “capitalize on leadership in vRAN, O‑RAN and AI‑RAN to capture new markets” hints that even its legacy networks know‑how may be repurposed for AI‑radio applications, further blurring the lines between telecom and AI hardware.
Uncertain Future for the Traditional Networks Division
Despite the clear strategic shift, Samsung’s traditional networks division remains a small and under‑emphasized part of its portfolio. The business earned only a fleeting mention on the Q1 earnings call, and no detailed roadmap was disclosed beyond the vague pledge to leverage vRAN, O‑RAN and AI‑RAN later in the year. Analysts warn that without a concrete plan to revitalize or possibly divest this segment, Samsung risks carrying a legacy drag that could impede the agility needed to compete in fast‑moving AI markets.
Conclusion: A Calculated Bet on Photonics, Memory, and AI
Samsung’s Q1 2026 performance illustrates a company at a crossroads. Declining revenues in its legacy networks business have forced a decisive reallocation of resources toward high‑growth optical and memory technologies that serve the AI boom. By announcing mass‑production of silicon photonics and CPO components, accelerating HBM shipments, and hinting at AI‑radio opportunities, Samsung is laying the groundwork to become a pivotal supplier in the AI‑data‑center ecosystem. If the execution matches the ambition, the conglomerate could transform a current weakness into a future strength—turning the tide from a faltering telecom vendor to a cornerstone of the AI infrastructure landscape.
https://www.fiercewireless.com/wireless/samsung-sees-light-ai-pivot