There’s a deal on the table to reopen the Strait of Hormuz, but Trump isn’t part of it—yet

Bill Gates warned the world is “unprepared for AI’s risks,” saying “there is no plan,” highlighting governance concerns around AI rollout. Separately, the article notes that AI-driven productivity gains are “refusing to show up,” a potential headwind for growth expectations. It also flags a deal aimed at reopening the Strait of Hormuz, but without clear signaling from Trump—keeping geopolitical risk sentiment mixed.

Analysis

This reads as a reminder that the AI trade is still being priced as an infrastructure capex cycle, not a verified productivity cycle. That keeps NVDA structurally supported in the near term because the buyer set is a handful of large balance sheets that can keep spending before end-demand proves out; the weaker link is the application/software layer that needs measurable ROI to defend its multiples. If productivity data keeps lagging, the second-order effect is slower enterprise rollout and more scrutiny on renewal/seat expansion, which is a 6-18 month problem for the broader AI software complex rather than an immediate problem for semis.

The main risk to the bullish infrastructure read is crowding: NVDA is now the default expression of the theme, so any pause in hyperscaler capex or export-control surprise can de-rate the whole trade quickly even if demand remains healthy. The immediate horizon is days to weeks of headline-driven volatility; the 1-3 month catalyst path is hyperscaler capex commentary and data-center build plans; the 6-18 month structural test is whether AI spend translates into operating leverage at the enterprise level. If that translation stays absent, multiple compression is more likely than a sudden earnings collapse.

The contrarian view is that the market may be overfocusing on the lack of measured productivity today. Productivity usually lags deployment by quarters, not weeks, so the absence of visible gains is not yet a thesis breaker for NVDA; it is, however, a warning sign that the eventual monetization curve may be flatter than current enthusiasm implies. That argues for owning the picks-and-shovels exposure more selectively than the broad AI basket, and for using any strength to trim exposure rather than chase it.

Key Decisions for Investors

• Prefer a hedged long NVDA expression over a naked AI-basket long: buy NVDA on 3-5% pullbacks and finance with short-dated upside calls if the stock is extended; thesis is capped if hyperscaler capex guidance rolls over.
• Pair trade: long NVDA / short XLK or IGV to isolate infrastructure demand against the risk that AI software monetization disappoints; target a 1-3 month catalyst window into next capex updates.
• If already long the AI theme via FVTI-like baskets, trim into strength and rotate toward semis only; this reduces exposure to the part of the trade most vulnerable to 'productivity not showing up' headlines.
• Set a watch trigger on MSFT/AMZN/GOOGL/META capex commentary: if any one of them signals slower data-center spend, reduce NVDA beta quickly because the re-rating could happen before earnings are revised.
• No aggressive short yet: wait for evidence of capex deceleration or multiple failure in NVDA before initiating a bearish position; absent that, the current signal is too weak for conviction shorting.