Since the battle for AI dominance erupted three and a half years ago, the four largest hyperscale cloud providers have collectively poured over $1 trillion in capital expenditures. Leading US tech giants are betting heavily on artificial intelligence, staking their future growth entirely on the technology.

According to earnings reports released by the four companies over the past two weeks, from the start of the AI industry boom in 2023 through the end of June this year, Alphabet, Amazon.com, Microsoft, and Meta Platforms, Inc. spent a combined $1.1 trillion on capital expenditures. This massive outlay highlights both the scale of their AI ambitions and a rapid shift by US tech giants from a light-asset model with limited investment to becoming major players in physical infrastructure. RBC Capital Markets analyst Rishi Jaluria noted, "The growth in capital expenditure currently seems to have no end in sight. Investors want these companies to strike a precise balance—continuing to invest in AI without harming the core businesses that have made them successful."

Both Alphabet and Amazon.com raised their full-year capital expenditure guidance in the most recent quarter. The four companies collectively plan to spend $745 billion on capital expenditures this year, with funds primarily directed toward data centers, advanced chips, and the power infrastructure needed to support their computing capacity. The success of this gamble partially depends on whether two AI startups, OpenAI and Anthropic, can continue to secure funding to honor long-term, multi-billion-dollar computing capacity purchase agreements. Both AI labs currently have plans to go public.

This concentrated wave of investment is putting pressure on supply chains, driving up hardware costs, and creating shortages in memory chips. Even Apple, which is not deeply involved in the AI arms race, has been affected. Apple warned that rising costs would lead to lower revenue and profit margins, causing its stock to fall 6.3% on Thursday. However, corporate earnings reports show that heavy investment is beginning to translate into accelerated revenue growth, particularly in the cloud computing sector. Cloud business revenue growth rates at Alphabet, Amazon.com, and Microsoft have all accelerated, driven by selling computing capacity to OpenAI, Anthropic, and traditional enterprises deploying AI. Strong cloud data boosted the stock prices of Amazon.com and Microsoft.

Meta Platforms, Inc. does not have a cloud business but reported that AI has effectively improved the precision of its ad targeting. The company's total revenue surged 28% year-over-year in the quarter, reaching $61 billion. CEO Mark Zuckerberg signaled plans to expand data center capacity for computing power rental services. He told investors that the company has received numerous requests to lease computing capacity, with tenants willing to pay a significant premium over Meta's own internal costs. However, he also cautioned, "The profit margins for selling intelligent services will be significantly higher than directly leasing out computing capacity over the long term." Dirk Muraki, Managing Director at asset manager SLC Management, believes that the vagueness of Meta's computing capacity rental plans was a key reason for its stock's 8% drop on Thursday following the earnings release. He commented, "Meta's overall strategy is messy and inconsistent."

"Investors are no longer accepting growth at any cost. They need to see investments translate into tangible results, and Alphabet, Microsoft, and Amazon.com have already set a precedent." While Alphabet's cloud business revenue increased by $11 billion year-over-year, the company reported its first negative free cash flow quarter in over two decades, with negative free cash flow of $6 billion. Investors still chose to sell the stock. Like its peers, Alphabet disclosed a massive expansion in its AI-related forward commitments, surging by approximately $500 billion compared to three months ago. Most of the new contracts are long-term procurement agreements for technology infrastructure and data center energy.

Meta Platforms, Inc. added $233 billion in various commitments this quarter: $96 billion in data center and network infrastructure leases (which will be capitalized upon facility activation), $112 billion in procurement agreements mainly for third-party cloud servers and supporting infrastructure, and $25 billion in new debt. In July, Meta added another $68 billion in data center lease agreements. Microsoft signed over $130 billion in new data center lease agreements in the second quarter, significantly expanding its forward commitments. In just one quarter, the three companies—Alphabet, Meta Platforms, Inc., and Microsoft—added nearly $900 billion in AI-related long-term obligations, deeply tying their future balance sheets to the AI arms race.

Several company executives acknowledged in analyst calls that AI investment will continue to erode free cash flow for the coming quarters. Free cash flow is a key metric closely watched by the market, representing the cash remaining after covering operating costs and capital expenditures, available for debt repayment and shareholder returns. The combined free cash flow of the four companies fell to a decade low of just $7 billion, with only Microsoft and Meta Platforms, Inc. showing positive cash flow from operations. Amazon.com CEO Andy Jassy told investors that the company needs to endure a period of free cash flow pressure. The companies are simultaneously building massive new data centers, which have a two-year construction cycle from groundbreaking to server installation and billing. "In the short term, we will maintain a high level of capital expenditure, and free cash flow will be under pressure until a large number of data centers are officially put into operation," he said.

The significant time lag between upfront investment in data centers and realizing revenue is evident. The broad sell-off in the AI sector over the past few weeks, driven by volatile investor sentiment, suggests that the market must be prepared to wait years before seeing substantial returns on these investments. RBC's Jaluria added, "Investors are now forced to extend their time horizon for expected returns."