Earnings analysis: AI costs rack up at Alphabet, Amazon, Meta and Microsoft

Capital expenditures (CapEx) across Google–parent Alphabet, Amazon, Meta and Microsoft are ramping up to fuel investments in artificial intelligence, which analysts predict may have a knock-on effect on advertising prices.

Meta raised its CapEx outlook for the year to between $35 billion and $40 billion—up from its prior estimate of $30 billion to $37 billion—to “support our AI roadmap,” chief financial officer Susan Li told investors last Wednesday. The company said it expects CapEx to continue increasing in 2025.

Li said additional infrastructure investments are required to fuel “more advanced and compute-intensive recommendation models and scale capacity for our generative AI training and inference needs.”

CapEx was $6.72 billion in the first quarter, a decrease from $7.09 billion the prior year.

Meanwhile, at larger Alphabet and Microsoft, spending levels ramped up significantly in the quarter and are set to remain high throughout the year.

Alphabet’s CapEx jumped 91% compared to the same period last year to $12 billion. CFO Ruth Porat said spending was “driven overwhelmingly” by investment in its technical infrastructure, including servers and data centres. She said investors can expect quarterly CapEx throughout the year to be “roughly at or above this level.”

Microsoft’s CapEx lifted 79% from a year earlier to $14 billion, with CFO Amy Hood attributing the rise to supporting cloud demand, including “the need to scale our AI infrastructure.” An investor pointed out that Microsoft is on track to ramp CapEx to over $50 billion this year, a 50% increase from 2023.

Across both companies, spend increases in the first quarter outpaced revenue growth. Alphabet’s sales grew 15% to $80.5 billion, while Microsoft’s sales climbed 17% to $61.9 billion.

On the other hand, Meta reported a record first-quarter revenue of $36.5 billion, up 27% from last year.

Despite strong growth, Meta’s stock tumbled 15% in after-hours trading, while shares at Alphabet and Microsoft rallied roughly 10% and 5%, respectively.

Brian Wieser, principal of Madison and Wall, attributed the negative reaction to Meta to analysts’ “unrealistic expectations” for growth for the rest of the year.

“These are really strong numbers, and they’re telling us that the whole industry is doing really well right now,” Wieser said.

Meanwhile, Amazon expects its CapEx this year to step up “meaningfully” from 2023’s $48.4 billion, chief financial officer Brian Olsavsky said on Tuesday, primarily to support its generative AI efforts. The company invested $14.9 billion in property and equipment in the latest quarter, up 5% from a year prior, which Olsavsky noted would be “the low” for the year.

Jacob Bourne, a technology analyst at Emarketer, suggested that tech firms may increase ad prices and platform fees or alter pricing models for AI-related services to recover the massive costs associated with AI infrastructure expansion.

“Particularly for Google and Meta, which depend so heavily on ad revenue, higher ad prices will likely be a strategy to offset the AI spending spree,” Bourne said.

He added that AI is also increasing efficiency in ad buying, which could lead to lower overhead costs for ad agencies.

Andrew Frank, the distinguished VP analyst in the Gartner Marketing Practice, echoed: “Ad prices and budgets are rising, but so is AI-driven efficiency within walled gardens. This can result in overall lower customer acquisition costs even as organisations spend more on media.”

The AI arms race is on

Analysts last week pushed Alphabet, Meta and Microsoft executives to explain why AI requires such high investment. Still, they could not share specifics beyond lofty goals of being seen as a “leader” in the field.

When asked by investors why Meta changed its investment outlook, CEO Mark Zuckerberg said the company has “gotten more optimistic and ambitious on AI,” adding that he believes it can “be the leading AI company in the world.”

Microsoft CEO Satya Nadella similarly discussed wanting “to be a leader in this big, generational shift and paradigm shift in technology,” while Alphabet’s Sundar Pichai talked about being “an AI-first company since 2016.”

Wieser, who formerly led business intelligence at media giant GroupM, said the companies need to be more transparent in their earnings calls.

Amazon president and CEO Andy Jassy was more forthcoming about the “tremendous opportunities” AI opens up for its cloud business, Amazon Web Services (AWS), whose chips AI developers like Anthropic are using to build out their models. Beyond training models, Jassy said “quite a few” companies are using AWS to make inferences and securely manage their data.

The e-commerce giant said its generative AI investments have already reached a “multibillion-dollar” revenue run rate and contributed to AWS’ 17.2% growth rate in the first quarter to $25 billion. Overall, Amazon’s net sales grew 13% to $143.3 billion, beating forecasts.

AI requires huge computing resources to run. According to Naveen Verma, professor of electrical and computer engineering at Princeton University, the amount of computing power required by AI models grew by about 1 million% between 2012 and 2022.

Market research firm Dell’Oro Group anticipates AI infrastructure spending will propel data centre CapEx to over $500 billion by 2027.

News of Alphabet, Meta and Microsoft’s growing AI investments boosted stocks of AI chip firms such as Nvidia, AMD, Broadcom and Marvell Technology.

In February, Nvidia CEO Jensen Huang predicted that data centre spending would double to $2 trillion over the next four or five years.