Since the start of the conflict in Iran, according to PBS, the average price of a gallon of gasoline in the US has increased by 46.3%. Here in Europe too, the ongoing disruption in the Strait of Hormuz – one of the world’s key energy shipping lanes – has seen the EU spend more than €30 billion on energy with no corresponding increase in supply.
At the same time, businesses and consumers are embracing the power of AI and the race to dominate AI has become a geostrategic concern for governments globally. Using the latest and greatest LLM on a laptop or phone can feel fairly disconnected from the physical world. The reality is however that AI is very much rooted in the physical world and is quickly becoming one of the most energy intensive industries there is.
According to Pew, in 2024 data centres in the US alone consumed 183 terawatt-hours (TWh) of electricity, 4% of the country’s total electricity consumption and equivalent to the annual electricity demand of the whole of Pakistan. And this figure is only going to grow, with data center energy usage expected to increase by 133% by 2030.
AI searches in other words, have a real world cost, consuming around 1000 times more electricity than conventional web searches.
So, with energy price inflation likely locked in for the rest of 2026 regardless of whether a resolution to the current conflict is found, what does this mean for the governments and businesses that are now so invested in AI?
The price of intelligence
Given the intense competition to lead globally on AI, not just for individual businesses but also for nations, there is little sign so far that the big players in AI plan on stepping back from data center investments.
As it stands, ecommerce giant Amazon is projecting data center spending worth $200 billion in 2026 alone with Google close behind on $185 billion and Microsoft plans $105 billion. When dealing with staggering sums of money like this, even a rounding error can be worth tens of millions, never mind an energy price increase of 50%. With a view to the long-term value of the AI market however, the tech giants are ploughing ahead.
The increased costs will need to be absorbed somewhere however. Tech researchers IDC forecast memory will cost $9.71 per gigabyte by the end of 2026, up from $3.76 in 2025. While some players have said that costs won’t be passed on to consumers, keeping the AI dream alive may eventually mean that the costs are passed on.
AI has become a key competitive edge for countless businesses and is central to the operations for many. The question for these businesses as we move further into 2026 will be how to manage the growing cost of intelligence?
FX – the hidden lever
For somebody who watches FX markets on a daily basis, it’s not every day we get to watch in real-time how the power of the dollar works. The energy price spike we’re currently seeing however presents such an opportunity.
While US consumers are undoubtedly suffering at the pump, the United States position as the biggest Oil producer in the world, has seen the value of the dollar rally 3% during the most intense period of the conflict as stronger growth due to petroleum exports and higher expectations of future inflation.
Crumbling consumer sentiment and rising inflation will also likely edge the Fed towards holding interest rates steady too, creating a cycle where the Dollar struggles to lose. It’s likely that, as economic concerns grow, the value of the dollar will increase as investors look for a safe haven. Consumers under pressure will also likely edge the Fed towards holding interest rates steady too, which again will increase the global attractiveness of the dollar.
And with oil being priced globally in dollars, rising oil prices also further increase the demand for dollars, pushing greenback values up and offsetting some of the impact of expensive energy.
The takeaway? American AI could become globally competitive as the energy price remains high, further undermining Europe’s own AI ecosystem. It’s not just the Fed that can play FX, businesses can too and those highly exposed to energy intensive AI would be wise to strategise this.
AI done smartly
Aside from reviewing where and how in your business AI is used to see if efficiencies can be found, there are a number of measures businesses can take to ride out the current energy price spike while continuing to lean into the efficiencies AI is creating.
While businesses might not have the heft of the Fed as a lever to pull, they can still adopt smart FX strategies to mitigate some of the risk. First up, businesses with an FX strategy should be considering forward FX contracts to lock in current prices and thus mitigate the risk of further dollar inflation devaluing currencies such as the Pound and the Euro.
This takes some of the unpredictability out of AI pricing and, with most commentators forecasting that energy prices will remain high over the medium term, it is likely a low-risk strategy.
When it comes to energy contracts too – where possible, businesses should agree fixed prices with their energy providers, mitigating the risk of further inflation.
Finally, with AI tools and services themselves, there’s no getting away from the fact that businesses on this side of the Atlantic simply don’t have a native AI sector that can compete with that of the US. So, ditching Silicon Valley for an alternative priced in Euros simply isn’t an option at scale.
That said however, businesses can introduce some diversification to their AI budgets and start backing some of those AI businesses trying to put the UK and the EU on the map and, as the energy price spike continues to bite, these businesses may come to look increasingly competitive over the coming months.
Hopes for an easy 2026 have been sunk in the Straits of Hormuz but at this stage at least, it seems unlikely that AI will be. The LLMs will keep iterating, whatever the costs. Given this, businesses embracing the ever-growing range of possibilities that AI offers should strategise now for a potential AI price spike and get ahead of this new form of digital inflation.
