PricewaterhouseCoopers: The wave of AI infrastructure investment is unprecedented, and global data center spending will reach $32 trillion by 2050

The Zhitong Finance App learned that the world is facing an unprecedented investment frenzy. The latest PwC report predicts that global data center spending will reach $31.6 trillion by 2050 to support the world's growing demand for artificial intelligence (AI). By comparison, the US GDP is around $30 trillion.

What's even more shocking is that PricewaterhouseCoopers pointed out in the first “Global Data Center Outlook” report released on Wednesday that if AI penetration exceeds its benchmark scenario forecast, data center spending may even reach 50 trillion US dollars in the next 25 years, far exceeding capital expenditure cycles such as railways, the Internet, and electrification.

As consumers, businesses, and governments increasingly use AI, tech giants such as Microsoft and Amazon, and many small to medium data center providers are rapidly deploying new computing power facilities around the world. Part of the expenditure will go to hardware equipment in the data center, such as Nvidia chips.

At the same time, the technology industry is also trying to calm the opposition against data centers, and this resistance is already threatening the construction process. According to statistics from the research institute Data Center Watch, in the first three months of this year, at least 75 data center projects around the world were shelved or delayed due to local protests, involving a total investment of about 130 billion US dollars. The main concerns of the protesters focused on environmental impact, resource consumption, and the deep impact AI could have on employment and social structures.

The wave of AI infrastructure investment is unprecedented, and US spending accounts for half

According to PricewaterhouseCoopers, the US will account for nearly half of the estimated estimated data center spending, reaching $15.1 trillion. The Asia-Pacific region is followed by $8.2 trillion, with Europe estimated at $5.6 trillion, and the Middle East and Africa at $1.1 trillion and $255 billion, respectively.

Unlike traditional infrastructure investment cycles — such as past memory chip production line construction or global optical fiber internet installation, which usually adopts the “early centralized investment” model — this round of AI infrastructure investment will continue to grow. Hardware such as GPUs, servers, storage systems, and network equipment needs to be replaced regularly, and continuous upgrading of computing power chips will form the main investment rather than land or building construction.

PricewaterhouseCoopers researchers wrote in the report: “Railways, electrification, and the internet all require huge amounts of capital and have defined an era. And the scale of the ongoing AI infrastructure construction cycle far exceeds those three. More importantly, this cycle resets every four to six years, and there's no sign of ending.”

Looking at the annual scale, PwC predicts that global data center spending will grow from about 800 billion US dollars this year to 1.1 trillion US dollars in 2030, and further rise to 1.8 trillion US dollars by 2050. The report specifically points out that China and India will be the biggest contributors to incremental demand, thanks to the two countries' huge population base, the rapidly expanding digital economy, and the huge application potential of AI in the commercial and consumer fields.

Key variables: electricity, chip supply chains, and sovereignty games

PricewaterhouseCoopers commissioned the Oxford Institute of Economics to model data center capital expenditure. The report covers 46 countries and five regions, which account for the majority of global economic activity and digital infrastructure investment.

PricewaterhouseCoopers stressed that despite strong global demand, power availability, data sovereignty requirements, and semiconductor circulation conditions will determine which regions can eventually capture these investments. Among them, power supply is the primary factor that determines the AI infrastructure investment layout.

According to the report, the forecast results are largely dependent on how quickly reliable power supplies for data centers are established. For many markets, large-scale provision of affordable, reliable, and increasingly low-carbon electricity is the most difficult requirement to meet.

Furthermore, the report assumes that the global trade system remains relatively open and chips circulate freely, but disruptions in the semiconductor supply chain may reduce global investment by nearly 20%. At the same time, rising claims for sovereignty will drive the redistribution of the investment pattern, but will not reduce the total amount of investment.

The researchers said, “The $31.6 trillion question is not whether the capital is in place — it's there; nor is it whether the demand is real — it actually exists. The real question is which regions, operators, and agencies can seize this opportunity and which will miss it.”