AI Infrastructure Week

The AI trade is taught as a GPU story. The data this week says the bottleneck has moved to optical. Cisco's $9 billion AI order book, Lumentum's 22 percent implied Q4 sequential acceleration, and a basket of optical-networking names quietly outperforming the chip names — the picks-and-shovels are no longer where most retail readers are looking.

The media declared photonics "phase 3" of AI infrastructure. The stocks just dropped 9–21% in a week. The physics didn't change — the price did.

Both have $2 billion Nvidia investments. Both make optical modules for AI. One is growing 90%. The other is growing 20%. The market treats them as the same trade — they're not.

37 of 39 analysts say Buy. Options are 89% calls. But the supply chain Nvidia is supposed to turbocharge just dropped 5–14% in five days. Someone is wrong — and we find out at 2 PM Pacific.

Nvidia beat for the 14th straight quarter and guided $91 billion. The stock barely moved. But the supply chain split in two — and the split reveals exactly where the market thinks the next constraint lives.

After Nvidia’s post-earnings week, the AI stack has sorted into five distinct temperature zones. Four layers have repriced. One refuses to move. That’s the next bottleneck.

The AI trade is no longer one stock. It is a stack — compute, networking, optics, memory, power, cooling, servers, and the grid. Five of those layers got paid this week. Three are still waiting.

Every watt that powers an AI chip becomes heat. That turns Vertiv, Eaton, Trane, and Carrier into part of the AI trade — even when the market is selling them.

The AI trade is no longer just about chips. It is now bounded by three physical limits — power, heat, and distance — and the eight layers of companies that solve them are the next dollar of AI capex.

Same physical constraint, four different economic models. The cooling layer is still flat-to-down 20 days into the AI capex confirmation. Here is the company-by-company decomposition — the day before Marvell prints and two days before Dell.

Five analyst firms raised price targets in May. The CFO and COO sold twelve days before the print. The stock is $34 above the mean PT. Marvell reports Q1 FY27 tonight at 4:45 PM Eastern — by tomorrow morning, one tape is right.

Marvell crushed Q1 on every line — $2.7B guide, FY28 raised to $16.5B, HSBC set a $300 PT — and the stock dropped 4.59%. Dell has run +48% in 20 days. RSI 80. Mean PT $240. Stock $305. Same physics. Dell reports after the close.

High-bandwidth memory is the scarce input the entire AI buildout runs on, and Micron is the cleanest U.S. way to own it. The market already knows: MU has run +79% in 20 days to $923 — about 50% above the $615 mean analyst target, RSI 76, insiders net-selling $45M. The thesis is real. The entry is stretched.