Brent crude oil surpassed $98 per barrel this Tuesday, a level not seen since October 2023. Gas stations are updating their prices upwards, European governments are preparing new transport subsidies, and in developing countries the cost of electricity and fuel is once again triggering social alarm. But what many analysts are now asking out loud is whether this escalation is not just about OPEC cuts or geopolitical tensions in the Middle East, but about a quieter and newer phenomenon: the unstoppable electricity demand of artificial intelligence.
According to the International Energy Agency (IEA), data centers dedicated to AI could consume between 85 and 134 additional terawatt-hours (TWh) per year by 2027, a figure comparable to the annual electricity consumption of countries like Sweden or the Netherlands.
When a click is worth more than a barrel
Every time a user interacts with a large language model like GPT-4 or Gemini, or every time a company trains a new computer vision system, servers turn on and work for hours or days. A single training run of an advanced model can consume as much electricity as an average household uses in several years. And although efficiency improves, the volume of use grows exponentially. Today there are more than 8,000 data centers operating worldwide, and their number is expected to double in the next five years. That means more natural gas burned to generate electricity, more coal in countries with dirty grids, and more pressure on the global fossil fuel market.
How is AI connected to oil prices?
Although most data centers run on grid electricity, a significant portion of that electricity is generated with natural gas and, to a lesser extent, oil. When AI's electricity demand pushes total energy consumption up, gas demand rises, which in turn makes crude more expensive in futures markets. Moreover, producer countries see an opportunity to keep prices high, knowing that global digitalization cannot do without firm power.
The geopolitical factor: between OPEC and Silicon Valley
OPEC and its allies, led by Saudi Arabia and Russia, have maintained production cuts since 2022 to support prices. But what was once a purely supply-side strategy now intersects with a demand growth nobody anticipated so strongly. According to Goldman Sachs data, 12% of the global electricity demand increase in 2025 came directly from the tech sector, and within that percentage, AI accounts for more than half. This has led countries like the United States, until recently net energy exporters, to see their export margin shrink because domestic gas and coal plants are feeding the servers of big tech companies.
Consequences already felt on the street
The rise in crude prices is directly passed on to gasoline, diesel, heating oil and plastics. In countries like Spain, a liter of gasoline is already close to 1.80 euros. In Argentina, fuel prices rose 15% in April. And in African oil-importing nations like Kenya or Senegal, the cost of generating electricity with diesel has soared, affecting hospitals, schools and small industries. The International Monetary Fund has already warned that if the barrel stays above $100 for several months, emerging economies could face a new debt crisis.
Are there alternatives to break the circle?
The solution is not to turn off artificial intelligence, but to rethink how it is powered. Large companies like Google, Microsoft and Amazon have promised 100% renewable data centers, but today only a fraction of their consumption comes from directly contracted clean sources. Nuclear, wind and solar power have limits of intermittency and storage. Meanwhile, governments are starting to demand transparency: a European directive in the pipeline will require tech companies to report the energy consumption of their AI models. Perhaps the key lies in more efficient algorithms and in regulation that puts a real price on the carbon generated by each query.
The debate has definitely changed tone. We no longer talk only about whether AI will replace jobs or bias decisions, but about how much energy it really costs to think for us. And as oil prices keep rising, that question will become increasingly political.