Satya Nadella believes cheap energy will device which country will win the AI race : Davos 2026

Microsoft CEO Satya Nadella just delivered one of the most important statements about the future of artificial intelligence, and it has nothing to do with better algorithms or smarter models. Speaking at the World Economic Forum in Davos, Nadella said that energy costs will be the deciding factor in which countries win the AI race. Countries with access to cheap and reliable electricity will have a structural advantage over everyone else, regardless of how good their technology is. This shifts the AI competition from a purely technical battle to one that depends heavily on infrastructure and resources.

AI tokens are now considered equivalent to oil

Nadella introduced the concept of AI tokens as a new global commodity, similar to how oil shaped international economics in the 20th century. In this context, tokens are the basic units of processing that users buy when they run AI models. Every time you ask ChatGPT a question, generate an image with DALL-E, or use any AI powered service, you are consuming tokens. Those tokens require computing power, and computing power requires massive amounts of electricity to run the data centers where all this processing happens.

The comparison to commodities like oil makes sense when you think about it. Just like countries with cheap oil had economic advantages during the industrial era, countries with cheap electricity will have advantages in the AI era. If you can produce tokens at a lower cost than your competitors, you can train bigger models, run more services, offer lower prices to customers, and ultimately capture more of the AI economy. Nadella put it simply: the job of every economy is to translate tokens into economic growth, and if you have that commodity at a lower cost, you win.

How much money is Microsoft planning to put into this AI race

The scale of investment required to compete in AI is staggering. Microsoft announced it expects to spend 80 billion dollars on AI data center construction. Half of that spending will happen outside the United States, which shows the company is looking globally for places where they can build infrastructure efficiently. That 80 billion dollar figure is not just for one data center or even a few. It represents a network of massive facilities that will consume electricity on the scale of entire cities.

To put this in perspective, training a single large language model with trillions of parameters can consume tens of thousands of megawatt hours of electricity. Running those models continuously to serve users requires data centers operating around the clock with advanced cooling systems and redundant power supplies. The operational costs do not stop after you build the facility. Every month, every year, these data centers are consuming enormous amounts of power, and the cost of that power directly impacts whether running AI services is profitable or not.

What are the major issues that Europe is facing

Nadella specifically called out Europe as being at a disadvantage because of high energy costs. Electricity prices in Europe jumped after Russia invaded Ukraine in 2022 and the subsequent sanctions disrupted energy supplies. While prices have come down from their peaks, European energy remains significantly more expensive than in many other parts of the world. This creates a real barrier to Europe competing effectively in AI development and deployment.

The Microsoft CEO pointed out that it is not just about producing energy but also about the total cost of ownership for running AI infrastructure. Can you build data centers efficiently? Do you have access to cheap, reliable power? Can you keep those facilities running without constant interruptions? Europe has been talking a lot about digital sovereignty and wanting to control its own technology destiny, but Nadella suggested that protecting European markets will not make the region competitive. The only thing that will make Europe competitive is if products coming out of Europe are globally competitive, and that requires dealing with the energy cost problem.

Then there is the issue of public acceptance

Nadella also raised an important point about public acceptance of AI energy consumption. He warned that society will quickly lose tolerance for using scarce energy resources to generate AI tokens if those tokens are not delivering clear benefits. The power that data centers consume could otherwise go to homes, hospitals, schools, and businesses. People will rightly ask why we are prioritizing AI when the technology is putting strain on electrical grids and potentially contributing to climate change through increased fossil fuel use.

This means AI companies and the countries hosting them need to demonstrate that the technology is actually improving lives in measurable ways. Better healthcare outcomes, more effective education, increased government efficiency, and enhanced private sector competitiveness across all industries. If AI just becomes a tool for generating viral images and chatbots that replace customer service with worse experiences, public support will evaporate. The social license to consume massive amounts of energy depends on AI delivering genuine value to society.

Which countries have the advantage in this AI race

The countries best positioned to win the AI race under this framework are those with abundant cheap energy and the ability to build infrastructure quickly. China generates twice as much electricity as the United States now, which gives them a cost effective foundation for expansive data center networks. They have been investing heavily in renewable energy and nuclear power to fuel their AI ambitions. If energy costs determine competitiveness, China has a significant structural advantage.

The United States still has relatively affordable energy compared to Europe, and American tech companies are pouring money into building AI infrastructure domestically. But there are growing political tensions around data centers in the US. Rising electricity rates linked to data center construction have become a political issue, especially in areas where utilities are struggling to meet surging demand. Some communities are pushing back against new data center construction because of concerns about grid stability and environmental impact.

Countries in the Middle East with access to cheap natural gas could also become AI hubs if they invest in the necessary infrastructure. The same goes for regions with abundant hydroelectric or geothermal power. Iceland, Norway, and other Nordic countries have attracted data centers for years because of their combination of cold climates that reduce cooling costs and cheap renewable electricity. Those same advantages apply even more strongly to AI workloads that require constant power and cooling.

The next few years will show whether Nadella is right that energy costs become the dominant factor in AI competitiveness. If he is correct, we should expect to see AI infrastructure concentrated in regions with the cheapest reliable power, while other areas struggle to justify the investment. That would reshape not just the tech industry but potentially the entire global economy as AI becomes more central to productivity and growth across every sector.