The author is a Reuters Breakingviews columnist. The opinions expressed are his own.

By Sebastian Pellejero

NEW YORK, June 9 (Reuters Breakingviews) - When AI is involved, no amount of money is too much. After Google parent Alphabet GOOGL.O clinched the initial part of an $85 billion stock sale last week, Facebook owner Meta Platforms META.O is now considering its own program, the Financial Times reported, with Microsoft MSFT.O and Amazon.com AMZN.O no doubt watching closely. All four have already tapped debt markets at record scale, as they race to build data centers to power the chatbot boom. With capital expenditures consistently outpacing Wall Street’s expectations, spreading across every funding source possible is now imperative.

At base is a vast spending spree. The so-called hyperscalers are collectively expected to dole out $676 billion in capital expenditures in 2026, according to Visible Alpha. That rises another 24% to $840 billion next year, peaking close to $1 trillion by 2030.

This is a dangerously voracious appetite: analysts expect next year’s spending to roughly match prodigious cashflow. Raising more cash is therefore a necessity, especially since the arms race only ever seems to run faster. Analysts expected about 20% spending growth for the large cloud computing providers at the start of 2024 and 2025, while actual growth exceeded 50% in both years, according to Goldman Sachs.

The group’s 21% annual growth in operating cash flow forecast through 2031 should just keep them in the black, but another unexpected spending spurt is always possible. As a result, each company is raising capital at a furious pace to maintain sufficient firepower. At this size, though, simply doing so through debt is a challenge. Their investment-grade issuance last year stood at over $73 bln, equivalent to about 5% of U.S. issuance, per PitchBook data. This year, investors expect closer to $200 billion, according to a Bank of America survey. Analysts note concerns that buyers may struggle to take down so much without hitting concentration limits.

Adding in equity therefore makes sense. This quickly runs into scale limits too, though. Last year saw only $123 billion of issuance by already-listed U.S. companies, according to LSEG. Alphabet's $45 billion haul thus far accounts for over 40% of secondary sales in 2026, reckons LSEG. Another $40 billion will be sold later in the year.

Rising costs may push capital hunger even further. Each generation of AI infrastructure costs more to build than the last, due to faster chips, denser networking, and higher energy requirements. Data centers from California to Virginia are waiting years to connect to the grid, prompting owners to secure their own power supplies. If spending keeps surprising upward, investors will need to keep their checkbooks handy.

Follow Sebastian Pellejero on LinkedIn.

CONTEXT NEWS

Meta Platforms is considering raising tens of billions of dollars in a stock offering to fund boss Mark Zuckerberg’s vast ambitions in AI, following Google’s record $85 billion share deal, the Financial Times reported on June 5 according to people familiar with the plans.

On June 8, Amazon.com launched a sale of roughly $10 billion of investment-grade bonds denominated in Canadian dollars, the largest corporate bond offering ever in the currency.