Moody's Ratings has quantified what many investors suspected but few could measure: the artificial intelligence buildout among the world's largest technology companies is now running on debt and lease obligations totaling well over a trillion dollars. A pension saver in London, a 401(k) holder in Ohio, and a fund manager in Frankfurt all share exposure to this story, because Microsoft, Amazon, Alphabet, Meta, Oracle and CoreWeave sit inside nearly every major index fund sold across the West.
The Capex Arithmetic Behind the Warning
The research note, released this week and dated Wednesday, projects that the six companies will spend $785 billion on capital expenditure in 2026, rising to about $1 trillion the following year. Moody's tracked approximately $460 billion in direct debt already sitting across the group's balance sheets, alongside $1.2 trillion in total lease commitments — a figure that dwarfs the direct borrowing and has drawn far less public attention.
More striking is the composition of those leases. More than $820 billion of the $1.2 trillion total relates to leases that have not yet started, meaning the financial obligation exists on paper well before a single data center comes online. That structure lets companies defer some costs off traditional balance sheets even as the underlying commitment locks in for years, a mechanic that credit analysts warn deserves closer scrutiny than it has received.
A Tale of Two Balance Sheets
Not every company in Moody's sample carries the same risk. Microsoft, Alphabet, Amazon and Meta retain what the ratings agency describes as the strongest corporate balance sheets in the group, giving them room to absorb capex swings without threatening their credit standing. Alphabet underscored that cushion last month with an $85 billion equity sale, a move that strengthens its position precisely as spending accelerates.
Oracle and CoreWeave sit in a different category entirely. Oracle carries a Baa2 rating with a negative outlook, which Moody's describes bluntly as sitting just two notches above junk status. CoreWeave operates within the high-yield market outright, relying on complex private debt structures to finance the GPU hardware fleets that underpin its business — a financing model that looks nothing like the conservative treasury operations traditionally associated with blue-chip technology names.
The Circular Ecosystem Problem
Moody's note also flags a circular ecosystem forming around AI labs including OpenAI and Anthropic, where compute providers, cloud infrastructure firms and the labs themselves are increasingly financially intertwined. When the customer, the supplier, and the financier of a technology buildout begin to overlap, the usual market signals that price risk accurately start to blur. That is not a new phenomenon in corporate finance history, but the scale here — a trillion dollars in projected annual spending within two years — is unprecedented for a single technology cycle.
Industry observers suggest the arrangement is simply how infrastructure gets built during any major technological shift, and that today's hyperscaler balance sheets remain far stronger than those of leveraged buyout targets from prior credit cycles. That view has merit for the four companies Moody's singles out as financially strongest, but it does little to address the exposure sitting with Oracle and CoreWeave, where the margin for error is visibly thinner.
What This Means for Western Portfolios
For US, UK and EU investors, the exposure runs deeper than direct shareholdings. Index funds, pension schemes and insurance portfolios across all three markets hold significant positions in these six companies, meaning a credit event at the weaker end of the group — Oracle or CoreWeave — could ripple through instruments that ordinary savers never chose individually. The private debt structures financing CoreWeave's GPU fleets are particularly opaque to retail investors, who have no direct visibility into how those obligations are structured or who ultimately bears the risk if hardware depreciates faster than expected.
That so much of this financing now flows through private credit and unstarted lease agreements, rather than transparent public bond markets, should concern anyone who remembers how quickly off-balance-sheet leverage turned into a systemic problem in previous credit crunches. Regulators on both sides of the Atlantic have said little publicly about the AI capex boom's financing structure, leaving rating agencies as the primary check on a spending pace that shows no sign of slowing.
The Next Test
The immediate marker for investors will be whether Oracle's outlook shifts further given its already-thin cushion above junk territory, and whether CoreWeave's reliance on private debt draws scrutiny as more GPU-backed financing deals come to market. With projected spending climbing from $785 billion this year toward roughly $1 trillion next year, the gap between the strongest and weakest balance sheets in this group is likely to widen before it narrows, and Western markets would be wise to watch which side of that gap each company falls on.
---
**Sources**
• [CNBC Finance](https://www.cnbc.com/2026/07/24/moodys-ai-spending-credit-quality-amazon-meta-alphabet.html)
*Disclosure: MorrowReport's publishing system uses Anthropic's Claude AI technology.*
