The Data Center Investment Boom Continues Despite 'Deepfake Shock'

AI Data Center Demand Set to Triple by 2026, as Major Tech Giants Continue Investment Surge / AFP

Even in the wake of the 'Deepfake Shock,' the frenzy of investment in artificial intelligence (AI) data centers remains unyielding. According to The Economist, Mark Zuckerberg, CEO of Meta, revealed plans on January 29th to construct a massive AI data center in Manhattan that would occupy a significant portion of the area. This announcement came at a time when the AI investment landscape faced skepticism, especially after a sharp drop in the stock prices of AI-related companies like NVIDIA and Dell—key players in AI chip and server manufacturing. Despite the initial drop, the stocks of these companies quickly recovered after Zuckerberg's announcement, signaling continued confidence in the AI sector.

This ongoing investment boom is not limited to Meta alone. Sundar Pichai, CEO of Alphabet (Google’s parent company), stated on February 4th that most of Alphabet's capital expenditure this year would be directed toward data centers, with an allocated $75 billion—significantly higher than the $53 billion spent last year. This indicates a broader trend where major tech companies are continuing their investments in AI infrastructure despite recent setbacks in the tech sector.

In 2024, the three major U.S. cloud service providers—Alphabet, Amazon, and Microsoft—invested $180 billion in data center infrastructure. When combined with investments from large data center operators like Digital Realty and Equinix, as well as smaller tech firms and telecom providers, the total investment reached $465 billion. Approximately 30% of this investment went into land, buildings, and electrical equipment, while 70% was directed toward chips, server racks, and networking kits.

The primary reason for this unwavering investment in data centers is the severe shortage of data center infrastructure worldwide. According to current estimates, there are approximately 11,000 data centers globally, consuming roughly 55 gigawatts (GW) of electricity. In terms of energy consumption, North America accounts for about 50%, Asia for 30%, and Europe, the Middle East, and Africa together consume the remaining 20%. In North America, the vacancy rate for data center space is a mere 2.8%, with lease contracts typically spanning 10 to 15 years. This limited availability makes it increasingly difficult to meet the rising demand for data storage and processing power.

Both Alphabet and Microsoft have cited capacity constraints as reasons for their slower-than-expected cloud business growth in the fourth quarter of 2024. As a result, many countries worldwide are accelerating their data center development to stay competitive in the global AI race. This has led to an increased demand for essential components, including transformers, which are critical for the operation of data centers. Some forecasts suggest that the waiting time for these components could extend to several years, causing delays in the construction of new facilities. According to John Lin, an executive at Equinix, the construction timeline for data centers has stretched from 12-18 months to as long as three years, further highlighting the persistent supply shortage.

At the same time, the demand for AI data centers continues to surge. According to Goldman Sachs, AI usage accounted for only about 10% of the total data center capacity at the end of 2024. This figure is expected to rise sharply as more companies adopt AI technologies. Goldman Sachs forecasts that the demand for AI data center space will triple by the end of 2026.

The Economist emphasizes that despite concerns over the potential slowdown in AI data center investments due to the Deepfake Shock, the demand for data center capacity far outstrips the available supply. As more businesses continue their digital transformation and migration to the cloud, the need for additional data centers becomes increasingly urgent. In this context, the ongoing investment in AI data centers is essential to meet the rapidly growing demands of AI-powered applications and cloud services.