AI Capex Doesn't Care About the Strait Today. Mid 2027 is a Different Story
The Nasdaq printed a fresh all time high above 29,300 today with the Strait of Hormuz still closed. Iran’s blockade is in its eleventh week. Qatari LNG Trains 4 and 6 are physically destroyed with a 3 to 5 year rebuild floor. Saudi, Iraqi, Kuwaiti, and UAE production was collectively 7.77 mb/d below the February print in March’s OPEC MOMR. Brent dated peaked at $144 on April 7 and has cooled to $113. The household consumer is breaking, with Home Depot 25% off its high and McDonald’s, Domino’s, and Wingstop all flagging the low income squeeze on Q1 calls.
And tech is ripping.
The crowd looking at this divergence and yelling that the strait is still closed is right about the closure and wrong about why it does not matter for tech today. Today is not the relevant horizon. The relevant horizon is 12 to 18 months from now, when the cascade reaches the data center bill of materials and hyperscaler dollars per megawatt of installed capacity step up 15 to 25%.
This piece walks through the lag mechanics, the specific input chains that hit the AI capex cycle, and most importantly what investors will actually be saying about all of this when the bill lands.
One thing to be clear on before walking through any of this. We are not bearish on the tech sector. The AI buildout is real. The compute demand is real. The structural growth in commoditiy demand is real, as is cloud, in chip design, foundational software, and in the platforms that monetize the new capacity is real. The argument here is narrower and more specific. It is about cost basis, ROI math, and the lag between an energy shock and the investment cycle that has to absorb it. The names that compound through the cascade are the ones with pricing power and balance sheet. The names that get hurt are the ones whose multiples assume input costs are stable and whose business models cannot pass through. The distinction matters, and the trade architecture at the bottom of this piece reflects it.