•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
After decades of operating as asset-light “money-printing” machines, Silicon Valley’s biggest tech groups are shifting into a high-spending, infrastructure-heavy phase. Record capital expenditure tied to artificial intelligence is pushing free cash flow to the lowest levels seen in about a decade, according to analyst compilations cited in the article.
Analysts compiled by Visible Alpha estimate that the four cloud infrastructure giants—Amazon, Alphabet, Microsoft and Meta—are pursuing an AI investment strategy totaling $725 billion. The spending is beginning to strain the companies’ cash generation.
In Q3 this year, total free cash flow for the group is expected to be only about $4 billion. That compares with a quarterly average of roughly $45 billion since the post-COVID period began six years ago.
For the full year, free cash flow for the group is projected to reach the lowest level since 2014, when their revenues were about one-seventh of today’s scale. The article describes this as a major turning point for companies that previously relied on an asset-light, cash-rich model while now funding large physical infrastructure buildouts.
Justin Post, an internet analyst at Bank of America, is cited as saying this is the deepest capital expenditure cycle the sector has ever witnessed. Despite the cash pressure, the article notes that chief executives are continuing to invest, framing the buildout as a once-in-a-generation opportunity to reshape the future.
Free cash flow is presented as a key metric for assessing a company’s ability to pay down debt or return cash to shareholders after operating costs and investments. This year, the article says the picture is deteriorating as the companies concentrate resources on AI.
Amazon is expected to spend more than it generates. Meta and Microsoft face cash shortfalls in at least one of the last quarters of the year. Even Alphabet, while still positive, is expected to see cash flow fall to a decade low.
The article describes the trade-offs as those typically associated with heavy industries: cutting staff, reducing shareholder payouts, or taking on more debt to fund ambitions.
Alphabet, for the first time since 2015, did not repurchase shares in Q1 2026. Instead, it issued $31 billion in new debt and recently raised an additional $17 billion through euro- and CAD-denominated bonds. CEO Sundar Pichai is quoted as saying the ability to invest now will help the company retain its leading position.
Meta has raised $55 billion in debt over the past six months and paused its share repurchase program, described as the longest pause since 2017. The article attributes part of the pressure to Meta’s lack of a cloud business that could lease out infrastructure, adding that Zuckerberg has cut staff to allocate resources to AI while acknowledging there is no detailed plan to recoup capital on a monthly basis.
Alongside the spending surge, analysts cited in the article express concern about accounting methods that may make financial results look stronger than underlying cash generation. The article says some conglomerates, including Meta, have pushed tens of billions of dollars in data center project costs off the balance sheet using special purpose entities (SPVs).
While the approach can help attract capital and borrow without fully showing liabilities on main financial statements, the article says it can also obscure who ultimately bears the risk if AI demand fails to meet expectations.
The article also notes that even Oracle, controlled by Larry Ellison, has used off-balance-sheet structures for a data center project valued at $300 billion for OpenAI. Oracle is described as having run negative cash flow since last year, with no expectation of returning to positive cash flow until 2030.
Christian Leuz, an accounting professor at the University of Chicago’s Booth School of Business, is cited as warning that the true free cash flow of many large companies may be worse than reported. He points to flexibility in calculating data center lease costs and stock-based compensation.
The article adds that hardware supply-chain pressures are increasing costs. Microsoft says component price inflation—such as memory chips—will push up its capital needs by $25 billion this year. It also states that the value of servers and networking equipment on Microsoft’s balance sheet has risen from $61 billion to $191 billion in just two years.
The article compares the AI spending boom to capital cycles in heavy industries such as telecommunications or chemicals, where over-investment can lead to excess capacity and low profits. However, it argues that in tech, “stopping is death.”
Leuz is quoted likening the dynamic to fear of missing out (FOMO): major tech players invest when rivals invest, driven by the fear of being left behind by a revolutionary technology. The article says this mindset reinforces and accelerates an unprecedented capital expenditure cycle, turning cash-rich companies into ongoing borrowers for a future with an uncertain end.
Premium gym chains are entering a “golden era” that is ending or already in decline, as rising operating costs collide with shifting consumer preferences toward more flexible, community-based ways to exercise. Long-term memberships are shrinking, margins are pressured by higher rents and facility expenses, and competition from smaller, more personalized…
