Arista Networks keeps outperforming because co-packaged optics is still years from scaling, while AI clusters need faster connectivity now through generations of pluggable optics first. Arista led in optimized switches and now leads in pluggables, as its management and others including Astera Labs and AMD said last quarter. It executes on the technology customers need when they need it during the transition to CPO. Silicom is doing the same. x.com 

Broadcom, Arista, Celestica, Credo and Astera Labs should inherit AI network spending as flatter data-center designs shift dollars from optical transceivers to silicon, integration and copper. B. Riley warns Amazon's RNG and OpenAI's MRC protocols could cut transceiver demand 40-50%, though the facility-wide drop looks nearer 25-35%. Broadcom ASICs run MRC routing, Arista enables flat clusters, Celestica builds custom racks, Credo copper replaces short optics, Astera retimers ease bottlenecks. … Show more

Prepaid fiber orders make 2027-28 demand a stronger signal than capex plans. Corning reports $639M in net customer deposits in the first nine months and expects data-center interconnect to beat its former $1B-by-2030 target on timing and size. That backs Lumen's intercity build, though its next allocation may require a bigger commitment. Tight Corning supply gives Amphenol room to gain share until capacity expands; strong demand alone has not lifted stocks across the chain. 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 

AI data-center spending flows beyond GPUs and operators into networking, optics, power and cooling. Nvidia, AMD, Broadcom and Marvell supply chips; Arista, Nokia and Credo connectivity; Applied Optoelectronics and Lumentum optics; Vistra, Constellation and Bloom power; Vertiv and Modine cooling. Meta's $140B spending this year illustrates the scale, per a commenter with whom the point is agreed. x.com  x.com 

Arista's newly announced rack-scale AI fabric adds a catalyst to the interconnect theme. Arista says its open Ethernet platform supports up to 144 accelerators per rack and cross-rack scaling; Arista joins Marvell and Semtech on the list. Lumentum remains listed despite its sale over a personal cost-basis display issue, with a possible repurchase after the wash-sale period. x.com 

Copper connectivity stays in Nvidia's racks longer than anyone modeled, a win for Amphenol and Credo. Per @SemiAnalysis_, the Kyber NVL144 rack is delayed more than 12 months to 2028 because its PCB midplane cannot be made at scale, and the NVL72x2 backup was cancelled. Co-packaged optics will not arrive until the Feynman generation, so the shift away from copper in scale-up moves out with it. x.com 

The shift from copper to light inside AI data centers is accelerating, and near-packaged optics is pulling optical demand forward. Lumentum's CEO said NPO demand looks bigger than CPO and arrived sooner, and he sees about $40 per share of fiscal 2028 earnings power, above sell-side estimates, with none of it in the numbers yet. Coherent's CEO said CPO and NPO engagement has jumped, with comparable content per socket. NVIDIA has moved its Spectrum-X CPO switch into full production. x.com 

Credo covers both short- and longer-range compute connectivity needs as those domains evolve. AECs should become even more critical for short-range scale-up and scale-out, while ALCs address longer-range scale-out. x.com 

Arista Networks benefits from the AI interconnect buildout because every transceiver has to plug into a switch. Switch port counts scale with accelerator count, and speed per port climbs on top of that. Arista sells to the handful of cloud titans doing the building, which make up most of its revenue. x.com