The global competition to build artificial intelligence infrastructure may not be decided solely by data center capacity or the availability of advanced processors, but by countries’ ability to provide sufficient, reliable, and reasonably priced electricity. According to PwC’s report, “Global Data Center Outlook 2026–2050,” global investments in data center buildings, servers, storage, networking, and advanced processing equipment are expected to reach $31.6 trillion between 2026 and 2050.
The report estimates that investments will reach approximately $800 billion in 2026, before rising to $1.1 trillion annually in 2030 and then to $1.8 trillion annually by 2050. These projections come as growing demand for artificial intelligence workloads increases pressure on existing electricity grids.
The Grid Is No Less Important Than Generation Plants
PwC places electricity at the forefront of the factors that will determine which countries and regions can attract large-scale data centers. The issue is not merely increasing energy production; available transmission-line capacity, access to substations, and the availability of transformers—which may take years to procure—could all delay project commissioning or determine their locations from the outset.
In practice, this means that building new generation capacity will not be sufficient if electricity transmission networks do not expand at the same pace to reach data center sites. The report indicates that some data center operators have begun developing dedicated generation capacity near their facilities, but these solutions may help meet the needs of individual projects and cannot replace the grid investments required for the sector’s large-scale growth.
Artificial Intelligence Increases Energy-Consumption Intensity
The growing pressure on electricity is linked to the nature of the hardware used for artificial intelligence workloads. Traditional cloud infrastructure relies more heavily on central processing units (CPUs), while artificial intelligence workloads use graphics processing units (GPUs) and accelerators that consume more power, in addition to being subject to faster technology-refresh cycles.
The report estimates that servers and GPU units typically need to be replaced every four to six years. The share of information and communications technology equipment in total data center investments is also expected to rise from 70% in 2026 to 93% in 2050. PwC estimates that every dollar spent on constructing a data center could subsequently drive approximately $12 in investments in information and communications technology equipment.
Renewable Energy and Data Sovereignty
The competition is not limited to the quantity of electricity, as providing competitively priced renewable energy gives some markets an additional advantage. The report cites Chile, owing to its strong solar resources, and Canada, owing to the stability of its grid and renewable-energy infrastructure, among the markets that may attract data center investments. In Europe, the Nordic countries may benefit from electricity systems that rely more heavily on renewable sources and from a climate that reduces cooling requirements.
Projections indicate that the Americas will account for $16.5 trillion in investments through 2050, including $15.1 trillion in the United States alone, or approximately 48% of the global total. The Asia-Pacific region is expected to receive $8.2 trillion over the same period.
Data-sovereignty policies could also redistribute investments. If governments and regulated sectors, such as finance, healthcare, and public services, move to keep sensitive data and artificial intelligence workloads within national borders, countries with strong domestic demand and currently limited data center hosting capacity could benefit. In this scenario, the report names Turkey and Poland as two European countries that could attract additional investment.
Why Does This News Matter?
The report’s significance lies in shifting the focus of the discussion from the race to purchase processors and build facilities to long-term operational constraints: electricity, its networks, its cost, and its source. However, the figures presented are projections, not announcements of confirmed investments, and the report does not determine which countries will actually overcome grid bottlenecks or how these projections will change as hardware efficiency and energy sources evolve.