Cloud titans Microsoft Corp., Alphabet Inc.’s Google and Meta Platforms Inc. are pouring money into AI – they hiked capex spending on the technology 60% year-over-year in the latest quarter – with plans to continue apace in 2025.
The trio is betting big as AI capacity becomes available and demand for general-purpose computing remains stout to support enterprises’ digital transformation plans to AI. At the same time, growth for cloud services providers improved in the last few quarters as enterprise IT budgets expand.
Results of S&P Global Ratings’ new report, “”U.S. Tech Earnings: Cloud Titans Bet Big On AI As Traditional Players Lag,” released Tuesday, highlight something else: Bifurcation in the U.S. tech economy.
A small number of large companies exposed to cloud and AI are outperforming the industry average, while companies exposed to mature markets like PCs and smartphones, or markets undergoing inventory corrections such as industrial, automotive and networking equipment, are underperforming.
“The major players are making substantial capital expenditure investments to support AI initiatives despite concerns about the timeline for monetization, S&P Global Ratings analysts concluded in the report.
But, they warned, “Overall, we believe the path to AI monetization and maturity will be longer than previously expected.” AI adoption by enterprises is still modest, they said, because large businesses are “still sorting through the proliferation of models and figuring out use cases.”
A tale of two industry segments is borne out in multibillion-dollar revenue run rates for Amazon.com Inc.’s AWS, Microsoft Azure and Google Cloud. Azure reported AI contributed 8 percentage points to its growth, an increase from 7 points the previous quarter. The number of Azure AI customers surged nearly 60% year-over-year and average spend per customer continues growing as usage expands across a wider range of models, according to S&P Global Ratings.
Google Chief Executive Sundar Pichai summarized the AI spending climate when he recently said the risk of underinvesting is dramatically greater than the risk of overinvesting. Meta said Llama 4 would take 10 times the compute to train as Llama 3. And Microsoft’s investments are for data center infrastructure with useful lives of at least 15 years.
While enterprises are delaying some non-critical, long-term projects, they generally have more confidence in the macroeconomic environment than six months ago, the study said.
“We expect a stronger second half of the year with signs of gradual recovery across most subsectors,” S&P Global Ratings analysts said. “Many segments are benefitting from the release of pent-up demand held over from 2023. The U.S. IT spending environment remains stronger than in Europe.”
