The Energy Squeeze Behind the Iran Conflict and the AI Surge

Key figures

• Brent crude rose to $104.3 per barrel on Tuesday, up 3.2% week‑over‑week.
• Global electricity demand from AI training workloads is projected to reach 350 TWh in 2025, a 45% increase from 2022 levels.
• The International Energy Agency (IEA) estimates the Middle East will supply 30% of the world’s oil in 2024, but recent attacks have cut output by an estimated 1.2 million barrels per day.

---

Business news

The confluence of two seemingly unrelated forces is reshaping the energy market. On one side, the United States and its allies have intensified air strikes against Iranian missile sites after Tehran’s recent escalation in the Red Sea. On the other, the rapid expansion of artificial‑intelligence training clusters—particularly in the United States, Europe, and East Asia—is driving a surge in electricity consumption that rivals traditional industrial loads.

Oil prices, which had been hovering near $90 per barrel for most of the spring, jumped to a three‑year high after the latest strikes disrupted the Strait of Hormuz. Simultaneously, cloud giants such as Amazon Web Services, Microsoft Azure, and Google Cloud announced plans to double their AI‑focused compute capacity by the end of 2025, a move that will require an estimated 150 GW of additional power, according to a recent analysis by the analyst firm Wood Mackenzie.

Market context

Geopolitical supply shock

Iran controls roughly 10% of global oil exports. The recent attacks have temporarily reduced its output by about 1.2 million barrels per day, according to the U.S. Energy Information Administration (EIA). While the disruption is expected to be short‑lived, the market reaction has been immediate: futures contracts for Brent and WTI have both surged, and inventories at Cushing, Oklahoma, fell by 3.4 million barrels in the past week.

AI‑driven electricity demand

Training large language models now consumes more power than the entire aviation sector in some regions. A 2023 study by the University of Massachusetts Amherst estimated that a single GPT‑4‑scale model can use up to 1.5 GWh of electricity during a full training run. As enterprises adopt generative AI for customer service, code generation, and analytics, the cumulative load is pushing national grids toward capacity limits.

In the United States, the Electric Reliability Council of Texas (ERCOT) reported a 6% increase in peak demand during the summer months of 2024, attributing a significant portion of the rise to data‑center consumption. European grids are seeing similar patterns; the Nordic power market recorded a record‑high demand for renewable‑backed electricity in June, driven largely by AI clusters in Sweden and Finland.

What it means for the industry

Utilities must diversify supply

Utilities that traditionally relied on baseload coal or nuclear generation are accelerating investments in flexible gas turbines and large‑scale battery storage to balance the intermittent nature of AI‑driven loads. For example, NextEra Energy announced a $4.2 billion expansion of its battery portfolio, targeting a 10 GW capacity addition by 2027.

Cloud providers face higher operating costs

The cost of electricity accounts for 30‑40% of total operating expenses for hyperscale data centers. With spot market prices for electricity in Texas reaching $0.18 per kWh during peak AI training windows, cloud providers are renegotiating power purchase agreements (PPAs) and exploring co‑location with renewable farms to lock in lower rates.

Investors are recalibrating risk models

Equity analysts at Morgan Stanley have raised the price targets for several oil majors, citing the short‑term price uplift from Middle‑East tensions. At the same time, they downgraded several utility stocks that lack flexible generation assets, warning that sustained AI‑driven demand could expose them to volatility in wholesale power markets.

Policy implications

Regulators in the European Union are considering energy‑intensity caps for AI training workloads, similar to the measures applied to cryptocurrency mining in 2022. If enacted, such caps could force AI developers to shift workloads to regions with cheaper, greener power, potentially reshaping the global distribution of AI compute.

---

Bottom line

The intersection of geopolitical conflict in the Middle East and the exponential growth of AI compute is creating a perfect storm for global energy markets. Oil prices are climbing due to supply disruptions, while electricity grids are being stretched by AI‑intensive data centers. Companies across the energy value chain—from producers to utilities to cloud providers—must adapt quickly, balancing short‑term cost pressures with longer‑term strategic investments in flexible generation and renewable integration.

Stakeholders who can secure stable, low‑cost power for AI workloads will gain a competitive edge, while those exposed to volatile fuel prices may see margins erode. The next six to twelve months will likely reveal how effectively the industry can align these divergent forces.