AI Capex Doesn't Care About the Strait Today. Mid 2027 Is a Different Story

Key Takeaways

• AI capex stays robust despite current oil supply disruption
• Power cost surge expected 12‑18 months after Strait closure
• Hyperscalers may pay 15‑25% more per megawatt installed
• Companies with pricing power and strong balance sheets will outpace peers
• Cost‑sensitive AI chip makers could face margin compression

Pulse Analysis

The closure of the Strait of Hormuz has driven Brent crude to swing from a $144 peak to the low $110s, cutting global oil supply by nearly 8 million barrels per day in March. While such a shock typically rattles energy‑intensive sectors, the AI build‑out has so far insulated tech firms because most of today’s spend is locked in before the price spike fully propagates. This decoupling explains why the Nasdaq can set new records even as households feel the pinch from higher fuel costs and reduced discretionary spending.

The real pressure point lies in the lag between an energy shock and the capital‑allocation cycle of data‑center operators. Power accounts for roughly 40 % of a modern AI‑focused facility’s OPEX, and as wholesale electricity prices climb, the cost per installed megawatt is projected to rise 15‑25 % within the next 12‑18 months. Hyperscalers will need to re‑evaluate ROI models, potentially delaying or re‑pricing new server farms. This timing aligns with the typical refresh cadence of AI‑optimized hardware, meaning the cost increase will hit just as the next wave of GPU and ASIC deployments is planned.

For investors, the distinction is clear: companies that can shift higher power costs to customers—cloud providers with entrenched contracts, chip designers with strong pricing leverage, and software platforms that monetize usage—are likely to maintain or expand margins. Conversely, firms whose earnings models assume stable utility rates—especially low‑margin AI chip manufacturers and smaller data‑center REITs—face compression risk. Monitoring balance‑sheet depth, pricing power, and contract structures will be crucial as the energy shock filters through the AI capex ecosystem.