AI Data Center Expansion Drives Trillions in Investment and Hardware Competition

Global spending on data centers is projected to reach US$ 31.6 trillion by 2050 to meet the rising demand for artificial intelligence, according to a report from PwC. This investment boom could potentially climb to US$ 50 trillion if AI adoption exceeds central forecasts. The United States is expected to account for nearly half of these expenditures at US$ 15.1 trillion, followed by the Asia Pacific region with US$ 8.2 trillion and Europe with US$ 5.6 trillion. The report notes that this infrastructure cycle is unprecedented, surpassing the historical scale of the internet, electrification, and railways. Much of the spending will be driven by the need to regularly replace hardware, such as graphics processing units (GPUs) and servers, every four to six years.

As the infrastructure race evolves, the focus is expanding from computing power to storage solutions. Samsung Electronics and SK hynix are currently competing for the enterprise solid state drive (eSSD) market to support AI inference workloads. According to TrendForce, combined revenue from the top five eSSD suppliers reached US$ 37.59 billion in the second quarter, a 103.6 percent increase over the previous quarter. Samsung currently leads this market with a 35.1 percent share, while SK hynix and its subsidiary Solidigm hold 21.1 percent. Both companies are focusing on high capacity quad level cell (QLC) technology to replace traditional hard disk drives in large scale AI environments.

While the financial outlook is expansive, some growth faces local resistance. Data Center Watch reports that at least 75 projects totaling US$ 130 billion were blocked or delayed in the first three months of this year. Opponents cite concerns regarding environmental impact, resource consumption, and the societal effects of AI on employment. PwC indicates that the availability of electric power will be a primary factor in determining where future AI infrastructure is located.