The AI Investment Wave: A Deeper Look

Dell’Oro analyst Baron Fung explains that the current growth isn’t simply about GPUs; it’s a comprehensive investment across the entire data center ecosystem. “AI compute costs are rising with increasingly sophisticated architectures,” Fung states. “We’re seeing investment in infrastructure, networking, storage, and even non-AI business areas.” This broad-based demand is creating a ripple effect throughout the technology sector.

The hyperscale giants – Amazon, Google, Meta, and Microsoft – are leading the charge, collectively increasing their data center capital expenditures by a staggering 76%. Recent earnings calls reveal no indication of a slowdown in this spending. Amazon, for example, invested $131 billion in capex in 2025, primarily for data centers, and anticipates spending approximately $200 billion in 2026, with the vast majority allocated to Amazon Web Services (AWS) due to exceptionally high demand.

The strength of demand is further evidenced by the growing backlogs at these hyperscalers. Amazon’s contracted future revenue now stands at $244 billion, a 40% year-over-year increase. CEO Andy Jassy emphasized the robust demand for AWS, both in the AI space and its core services. Google is similarly committed, planning around $180 billion in capex for 2026 and reporting a backlog of $240 billion. According to CEO Sundar Pichai, the number of deals exceeding $1 billion in value more than doubled in 2025 compared to the previous three years combined.

But what’s driving this unprecedented investment? The answer is unequivocally AI. AI developers require immense compute power to train increasingly complex models, while businesses deploying AI solutions are fueling demand for inference capabilities. Industry surveys consistently demonstrate a strong commitment to increased AI spending. A recent Boston Consulting Group survey of nearly 2,400 executives revealed that companies plan to double their AI investment this year, increasing allocation from 0.8% to 1.7% of revenues. Remarkably, over 90% of CEOs intend to maintain or increase AI investment levels, even if immediate returns aren’t realized.

The Rising Costs of Capacity: A Challenge for Enterprises

This surge in demand, however, isn’t without its challenges. The escalating investment in data center capacity is creating headwinds for enterprises seeking to deploy their own infrastructure. Hardware prices are climbing, particularly for memory, which can represent up to half the total cost of a server. Hyperscalers, with their significant purchasing power, are better positioned to absorb these increased commodity costs, but smaller companies and enterprises are feeling the pinch.

“Hardware prices are going up,” Fung confirms. “The cost of memory has gone up by double digits.” This is forcing some organizations to delay server purchases or extend the lifespan of existing equipment. Consequently, many are considering a greater reliance on cloud infrastructure rather than on-premise solutions.

There’s even speculation, Fung notes, that hyperscalers may be strategically acquiring memory to drive up prices and encourage greater cloud adoption. While this remains a hypothesis, it highlights the competitive dynamics at play. Regardless, Fung suggests that developing AI applications in the cloud initially is a prudent approach. “Before investing in capital spending, develop in the cloud first and test your AI cloud usage to see if you can really utilize your AI hardware constantly,” he advises. “Any idle time means you’re not getting the desired returns.”

What long-term strategies will enterprises employ to navigate these rising costs and maintain a competitive edge in the AI era? And how will the hyperscalers balance their own infrastructure needs with the demands of their enterprise customers?

Frequently Asked Questions About Data Center Investment