The landscape of artificial intelligence (AI) is undergoing a significant transformation, driven by a surge in demand for networking solutions that can support the infrastructure needed for advanced AI applications. As hyperscalers expand their AI clusters, the challenges they face are no longer limited to raw computing power but encompass issues related to bandwidth, latency, and the reliability of networking interconnects.
These AI clusters, often comprising over 100,000 compute nodes—including GPUs, accelerators, and CPUs—require substantial investment in optical modules to facilitate efficient data transfer among components like GPUs, switches, and server racks. The transition from traditional copper-based networking has prompted hyperscalers to adopt 800-gigabit (800G) optical modules at scale, with the use of 1.6-terabit (1.6T) silicon photonics-based interconnects moving into the advanced qualification and early adoption phases.
Despite the promising potential within the AI networking sector, investors have largely concentrated on GPU technologies, overlooking significant opportunities within the networking domain. Two companies poised to capitalize on this emerging trend by 2026 are Broadcom and Coherent.
Broadcom is experiencing substantial growth in demand for its AI networking components as data center infrastructure ramps up to support custom accelerator deployments. The company exited its fiscal 2025 with an impressive AI switch backlog exceeding $10 billion, primarily driven by pioneering bookings for its 102-terabit-per-second Tomahawk 6 switch. Broadcom has also received record orders for essential networking components, including digital signal processors, lasers, and PCI Express switches. Their total AI-related backlog stands at $73 billion, accounting for nearly half of a consolidated backlog of $162 billion. Notably, about $20 billion of this backlog is attributed to networking and optical components, underscoring their potential as a significant revenue source.
Broadcom’s custom accelerator business remains its core growth engine, with AI semiconductor revenue soaring 74% year-over-year to reach $6.5 billion in a recent quarter. Management anticipates that AI revenue will double year-over-year, projecting $8.2 billion for the current quarter. With its robust portfolio of accelerators and networking solutions, Broadcom is well-equipped to capitalize on the expanding global AI infrastructure market.
Coherent, on the other hand, specializes in designing and selling lasers, optical transceivers, and photonic components crucial for high-speed and low-latency data transmission in AI data centers, cloud networks, and telecommunication infrastructures. The company has reported a significant increase in revenue, with a 17% year-over-year rise to $1.58 billion in the first quarter of fiscal 2026. Notably, its data center revenue climbed 23% year-over-year, spurred by strong demand for high-speed optical connectivity equipment from hyperscalers.
Coherent’s record bookings signal promising revenue visibility and protection against macroeconomic volatility, with several orders stretching beyond 12 months. The company is also advancing the development of 1.6T optical transceivers utilizing multiple technologies, including silicon photonics and electro-absorption modulated laser. As customers increasingly scale their AI networks into 2026, broader adoption of these technologies is anticipated.
With the push for 800G deployments, initial ramp-ups of 1.6T solutions, and expanding order backlogs, Coherent is poised for rapid growth in the evolving AI networking landscape. Together, Broadcom and Coherent illustrate the significant opportunities arising from the integration of networking solutions within the burgeoning AI sector, highlighting a shift towards a more interconnected future in technology.
