Moody's Warns the AI Spending Boom Is Straining Big Tech's Balance Sheets

Moody's Ratings put a number on something that's been obvious to anyone watching Big Tech's earnings calls: the AI buildout is starting to reshape the financial DNA of the companies leading it. In a research note released this week, the agency warned that the scale of spending on data centers, chips and power is eroding free cash flow and pushing hyperscalers toward debt and stock sales in ways that threaten the credit quality of six major players — Microsoft, Amazon, Alphabet, Meta, Oracle and CoreWeave.

The core issue is a business-model shift Moody's says is "unprecedented." For most of the past two decades, the biggest tech companies ran what the industry calls asset-light operations — software and cloud services that scale cheaply once built, throwing off enormous margins and letting companies pile up cash rather than debt. That formula built some of the most fortified balance sheets in corporate history.

AI has broken that pattern. Training and running large models requires physical infrastructure at a scale software never did: data centers packed with power-hungry servers and specialized chips, all of it expensive to build and slow to pay for itself. Moody's described the shift bluntly, saying the move from asset-light to asset-heavy models "requires unprecedented levels of investment and capital raising."

The dollar figures back that up. Moody's projects hyperscaler capital expenditures — spending on physical assets like data centers, as opposed to operating costs — will hit $785 billion in 2026 and climb to roughly $1 trillion in 2027. Financing that kind of buildout is pulling tech giants deeper into Wall Street's orbit, which Moody's notes has been a boon for the finance industry even as it changes the risk profile of the companies footing the bill.

Direct debt across the six companies has already reached about $460 billion, according to Moody's. Companies are also raising cash through public markets — Alphabet, for instance, announced an $85 billion equity sale last month.

The off-balance-sheet workaround

Perhaps the more striking detail in the report involves how hyperscalers are trying to avoid loading up their balance sheets with conventional debt. Instead, many are leasing data centers through long-term agreements rather than owning them outright, keeping those commitments off traditional balance-sheet debt figures.

Moody's says those lease commitments have swelled to $1.2 trillion across the group, and more than $820 billion of that is tied to leases that haven't even started yet — meaning the data centers backing them are still under construction. The agency treats these leases as debt-equivalent liabilities, since they lock companies into major rent payments regardless of how AI demand actually plays out. In other words, the debt is real even if it doesn't show up where debt usually shows up.

Who's actually at risk

Despite the warning, Moody's isn't predicting downgrades for the industry's giants anytime soon. Microsoft, Alphabet, Amazon and Meta still carry some of the strongest corporate balance sheets on the planet, and the agency said their investment-grade ratings aren't under imminent threat, even as free cash flow tightens and borrowing headroom shrinks.

The real pressure point is further down the credit ladder. Oracle carries a Baa2 rating with a negative outlook — just two notches above junk status. CoreWeave, the specialized AI cloud provider that's become a poster child for the boom, operates in the high-yield market entirely, rated Ba3, and depends on complicated private debt arrangements to finance its fleets of GPUs. Those are the companies with the least cushion if AI revenue growth stalls or spending assumptions prove too optimistic.

A circular ecosystem

Moody's also flagged something structural that's easy to miss in the topline spending numbers: a lot of the demand hyperscalers point to is intertwined with their own investments. The agency noted that some of the massive backlogs companies report stem from deals with pre-IPO AI labs like OpenAI and Anthropic — labs that the hyperscalers themselves have invested billions into, and which then turn around and spend heavily on cloud computing from those same companies.

That circularity means many of the industry's largest players are leaning on the same customers and the same bets about future AI demand. If those assumptions turn out wrong, the exposure isn't isolated to one company — it ripples across several at once.

Moody's was careful to note the counterweights. Demand for AI computing remains strong, cloud businesses are still growing, and hyperscalers have locked in hundreds of billions of dollars in long-term customer contracts that provide a degree of revenue predictability. Those factors are propping up credit profiles across the industry even as spending balloons.

But the agency's broader message is that the financial character of these companies is changing in a way the cloud era never required. As Moody's put it, investors "will increasingly focus on these companies' ability to realize an adequate return on investment" — a shift in scrutiny that marks a departure from an era when Big Tech's cash generation was simply assumed.

What happens next likely hinges on whether AI revenue growth catches up to the capital being poured into it. Until it does, the gap between what these companies are spending and what they're earning back is the number worth watching.