Cisco is capturing the networking socket in AI systems built outside Nvidia's ecosystem. Saudi HUMAIN's AMD MI355X deployment links GPUs within nodes via AMD Infinity Fabric, but connects nodes and racks with Cisco Silicon One Ethernet and 800G pluggable optics, so Cisco's AI Ethernet stack carries real traffic and revenue in 2026. If this becomes a common sovereign-AI template, it could be quite meaningful. As Nvidia integrates vertically, non-Nvidia buyers need others for those layers. x.com 

More lasers per module and new 400G DSPs could keep the pluggable-optics runway open for years, supporting Lumentum, Coherent, Marvell, Applied Optoelectronics and Sivers even as co-packaged optics approaches. NVIDIA and Cisco co-designed a module with four times the bandwidth in the same slot; fewer modules per GPU look bearish, but the added optical content points the other way. x.com 

Inference's move toward edge devices and dense, memory-hungry racks should drive demand for specialized chips, links, power delivery and cooling. McKinsey projects inference at 60% of AI computing demand by 2030. Ambarella makes on-device AI chips; Astera Labs links GPUs, memory and networks; Credo supplies short copper connections; Rambus earns from memory interfaces; Navitas reduces power waste; Modine cools dense racks; and Super Micro builds inference-specific server racks at volume. x.com  … Show more

Higher optical-fiber use per GPU could expand demand for Applied Optoelectronics. Corning said fully optical scale-up would raise fiber count from about 16 to 160 per GPU; that 10x case is hypothetical for now, with a mix of technologies likely near term, but it shows the direction of travel. x.com  x.com 

Kopin, with partner SBLX, may have cracked MicroLED-based optical interconnects, the holy grail for interconnects, which are the main bottleneck in AI data centers. No other public company has announced one, and Kopin builds on MicroLED products it developed for the US Army. Copper-interconnect leader Credo is worth about $20 billion and just paid $750M for DustPhotonics, while Kopin is about $500M and SBLX under $50M in what is described as a $100B industry. x.com  x.com 

Credo's interconnect opportunity grows as campus-scale AI clusters require more links to validate and integrate. The OCP 2026 meeting takeaway is that complex requirements put interconnect at the center, with both copper and optical links set to benefit. x.com 

Optical networking's broader rollout will be slow because scale-up and scale-out require orders of magnitude more connections than today's limited optical distance links. Supply-chain discussions echo that slow pace. NPO/CPO starts in switches and reaches XPUs and GPUs much later, with new challenges. Optical backplanes will be necessary as racks move to optical, though copper will remain in many implementations. x.com  x.com  x.com  x.com 

Corning's optics business looks ready for a comeback as its fiber and high-density optical connectivity move data across massive AI GPU clusters. Optical sales rose 32% year over year to $2.07 billion, EPS is expected to more than triple by 2028, and the business is backed by major AI infrastructure deals with NVIDIA, Meta and Amazon. x.com 

Credo is a solid company whose low-power interconnects between AI servers become more necessary as AI compute grows. If microLED technology actually lands and improves optical module efficiency and density, its moat could get stronger. Filling the chart gap, or even a pullback to the 200-day moving average, would be a good buying opportunity, though Astera Labs was preferred when the two were compared. x.com  x.com 

Marvell looks well positioned for scale-up compute domains, where interconnect is the wall that limits performance. Scale-up domains should be thought of differently from simply networking racks together, and Marvell's own commentary now focuses on exactly that opportunity. x.com  x.com