Despite Bubble Fears, Big Tech Data Center Spending Is Accelerating Faster Than Ever

The world’s largest tech companies are doubling down on their massive investments in AI infrastructure and pushing back on claims that this data center spending spree is creating a bubble that threatens the global economy.

Wall Street may be getting nervous about Big Tech’s artificial intelligence spending, but Amazon, Microsoft, Google and Meta are ramping up their investments in new data centers, chips and other computing equipment even faster than expected. Capital expenditures from the four major hyperscalers jumped more than 18% last quarter from the three months prior.

The $113.4B in combined capex represented more than a 73% year-over-year increase. 

While the quarter marked an acceleration of what was already an unprecedented surge of digital infrastructure spending to support AI, the tech giants plan to press the capex gas pedal even harder in the months ahead. Their combined projected spending for 2025 is approaching $400B — tens of billions of dollars above what they predicted just months ago — and is expected to increase meaningfully in 2026.

Over the next two years, Amazon and Microsoft plan to double their data center footprints.

Tech behemoths are going all-in on scaling up their AI computing capacity even as rumblings of a looming AI bubble have grown louder. There is growing anxiety among some investors, economists and industry observers that spending on data centers and chips, which now accounts for more than half of U.S. GDP growth, could amount to a hype-driven boondoggle that never produces promised returns. Some have compared this moment to the days just before the dot-com crash of 2000.

Bubble fears were top of mind for the leaders of the big four tech firms on quarterly earnings calls last week. Amazon CEO Andy Jassy, Meta founder Mark Zuckerberg and other executives sought to reassure investors that their firms’ escalating capex isn't a reckless spending spree but rather cautious capital deployment that is driving revenue. 

Far from the kind of speculative infrastructure build-out that characterized the later days of the dot-com bubble, they said this scale of spending is necessary just to catch up with existing customer demand for computing capacity. 

“We know we're behind. We do need to spend,” Microsoft Chief Financial Officer Amy Hood said Wednesday. “Our need to continue to build out the infrastructure is very high, and that's for booked business today.”

Microsoft previously said its AI infrastructure spending would slow in the current fiscal year. Instead, its capex reached a record $34.9B last quarter, $5B more than the company’s prior projections and a 75% jump from the same quarter last year.

Now, the firm expects an even higher growth rate in 2026, with plans to increase total AI capacity by 80% in its current fiscal year and double its data center footprint over the next two years. 

Similarly, Alphabet's Q3 capex rose 7% sequentially to $24B and 85% year-over-year. The company now projects annual spending up to $93B, an $18B increase from earlier forecasts.

Meta also raised its annual capex guidance by $4B after a quarter with $19.4B in spending, up 14% sequentially and 111% year-over-year.

Amazon added a gigawatt of data center capacity last quarter, bringing its total added over the past 12 months to 3.8 gigawatts, more than any other cloud provider, Jassy said Thursday. The company has doubled its cloud infrastructure capacity since 2022 and expects to double it again by 2027.

That expansion means a dramatic increase in capex: a record $35.1B, up 9.3% over the previous quarter and 55% year-over-year. 

“You're going to see us continue to be very aggressive in investing in capacity because we see the demand,” Jassy said.

Other tech leaders last week also attributed accelerated capex growth to existing demand, rather than the anticipation of it, as they looked to address fears of an AI bubble.

Historically, tech gold rushes have led to overspending on infrastructure, causing speculative bubbles. The dot-com boom saw a massive build-out of fiber and data center capacity in anticipation of demand that took decades longer to materialize than investors thought, triggering a wave of bankruptcies. 

So far, public markets have tolerated Big Tech firms pushing well beyond traditionally accepted limits with their infrastructure spending — ramping up capital expenditures to as much as 50% of income. But there are signs that Wall Street is getting skittish about the scale of these investments as AI revenues lag behind these costs, with investors searching for signs of speculative, wasteful or short-sighted spending on data centers and chips. 

Tech leaders last week differentiated their soaring data center and chip costs from the fiber boom of the dot-com era. They said they are unable to keep up with surging demand for AI computing, products and services due to infrastructure shortages. 

Hood said Microsoft’s top-line revenue growth was artificially suppressed because a shortage of data center capacity delayed the execution of cloud computing deals until more data centers came online. She said the firm is accelerating capex because demand for AI products and services materialized faster than expected.

“I thought we were going to catch up,” Hood said. “We are not. Demand is increasing.”

Still, even if overall demand for AI computing continues its stratospheric growth trajectory, there are concerns that the AI ecosystem might evolve in ways that could render newly built data center inventory obsolete. 

A growing share of demand stemming from AI inference raises questions about the long-term viability of remote data centers, which are primarily used for training large language models. Doubts are further fueled by the release of more efficient AI models, leading to uncertainty about hyperscalers' committed infrastructure investments.

Or what if the AI startups with whom hyperscalers are signing billion-dollar cloud contracts prove to be short-lived? In a nascent industry experiencing such a massive flood of investment, this uncertainty about AI’s future embeds risk in Big Tech’s data center spending that has caught investors' attention. 

Microsoft CEO Satya Nadella said the company is mitigating this risk by building data centers that are sited and designed for a wide range of use cases to ensure that the investment isn’t wasted if the AI demand landscape doesn’t evolve exactly as envisioned. 

“We are building a fungible fleet that's been continuously modernized and spans all stages of the AI life cycle, from pretraining to posttraining to synthetic data generation and inference,” Nadella said. “It's not like we're building one data center in one region in the world that's megascale.”

Tech leaders also looked to head off the bubble narrative by showing they are monetizing AI and that their data center and chip spending is driving revenue growth, if not profitability. 

Amazon, Google and Microsoft all pointed to growth of their cloud segments as being driven in part by AI-specific demand from customers. They also outlined revenue from the AI products and services they offer, like AI agents, AI search engines and AI integrations that improve profitability in existing tools and product lines, like advertising or recommendation engines. 

Alphabet CEO Sundar Pichai led off the company’s earnings presentation Wednesday by touting the impact of its massive infrastructure investments on its bottom line.

“We are seeing AI now driving real business results across the company,” Pichai said, adding that Alphabet delivered its first quarter with $100B in revenue. “Five years ago, our quarterly revenue was at $50B. Our revenue number has doubled since then, and we are firmly in the generative AI era.”