For most of the past decade, Big Tech’s defining trait was not innovation but liquidity. Google, Meta and Microsoft ran fortress balance sheets with net cash positions in the tens of billions, funding themselves internally while barely touching the debt markets. That era is ending, and quickly. The capital appetite of artificial intelligence has pushed the world’s largest technology companies into the bond market at a pace few would have predicted even two years ago. Amazon is the exception, having never truly been asset-light to begin with.
From Fortress Balance Sheets to Sovereign-Scale Borrowing
As recently as 2024, Alphabet, Meta and Microsoft could all comfortably call themselves net cash companies. That is no longer true. Meta crossed into net debt for the first time in its history during FY2025, after its long-term borrowings more than doubled in a single year, from $28.8 billion to $58.7 billion. Alphabet has not yet crossed that line, but its gross debt has more than doubled since the end of 2024, as it tapped bond markets in one currency after another. The direction of travel is the same for all three: balance sheets built for capital discipline are being redeployed for capital intensity.

The CapEx Behind the Debt
What is driving all of this borrowing is not subtle. Capital expenditure at the hyperscalers has grown to levels that would have seemed implausible five years ago, and it shows no sign of plateauing. Combined capex across the five companies in this analysis is expected to approach $600 billion in FY2026 and climb toward $672 billion the year after. For some of these businesses, spending is now running ahead of the cash they generate from operations, meaning the shortfall has to come from somewhere else. Alphabet’s free cash flow in the first quarter of 2026 fell by roughly 47% year-over-year, to $10.1 billion, as infrastructure spending front-loaded hard.

Wall Street’s New Favourite Borrower
The bond market has stepped in to fill the gap, and at a scale that is beginning to reshape the investment-grade universe itself. High-grade technology issuers sold more than $200 billion of debt in 2025 alone, and some estimates put total AI-related borrowing needs as high as $1.5 trillion by 2028. Meta’s $30 billion bond sale in October 2025 was reportedly the largest US high-grade corporate offering of the year, drawing an extraordinary $125 billion in investor orders. Alphabet followed with roughly $32 billion raised across global markets in February 2026, including a 100-year sterling bond, apparently the first of its kind from a technology company since the 1990s, before returning in May for its largest-ever euro issuance and a debut yen deal. Even Nvidia, not a hyperscaler in the traditional sense but central to the entire buildout, sold $25 billion of high-grade bonds in June 2026 to $85 billion of demand. By the fourth quarter of 2025, the largest hyperscalers had collectively flipped from being net recipients of interest income to net payers, a symbolic as much as a financial turning point.
Amazon: Different From the Start
Amazon’s story runs parallel to this narrative rather than through it. Its logistics and fulfilment network has always demanded heavy capital investment, long before AI became a line item, which is why its balance sheet was never asset-light to begin with. Its gross debt already stood above $100 billion back in 2020. What has changed is the scale: Amazon expects to spend around $200 billion on capex in FY2026, over 50% more than the prior year, driven largely by AWS and AI infrastructure. The company’s advantage is that it pays out little in dividends or buybacks, leaving virtually all of its operating cash flow available to fund that spending, which softens, without eliminating, its need to tap external markets.
What This Means for Credit Markets
None of this is going unnoticed by credit investors. Warnings have already surfaced on both sides of the Atlantic that a continued flood of mega bond offerings could overwhelm buyers and strain the market more broadly. Meta’s credit default swap spreads hit a fresh high in April 2026 after the company flagged a further potential $25 billion bond sale, and it is worth noting that derivatives on Meta’s debt only began trading actively in late 2025, itself a sign of how quickly its credit profile has shifted. There is also a less visible layer of risk building up off balance sheet: data-center lease commitments across US tech companies topped $850 billion in the first quarter of 2026, up 204% year-over-year, obligations that sit alongside, rather than instead of, the debt already discussed.
Conclusion
The AI buildout has rewritten Big Tech’s financial identity. Meta and Microsoft have lost their net cash status; Alphabet’s is eroding fast; Amazon is simply accelerating a trajectory it was already on. For portfolios exposed to these names, through equity, Magnificent 7 concentration, or investment-grade credit, this deserves attention: balance sheet strength was long part of the investment case, and that pillar is now being traded for AI returns not yet visible in the numbers. Not a reason to exit the theme, but to stay disciplined, watch free cash flow, capex-to-revenue, and credit spreads, and be mindful of concentration risk. Whether this capital is well spent will only be clear in a few years. Until then, caution seems warranted.
Data sourced from Bloomberg Terminal. CapEx and balance sheet figures reflect most recently reported fiscal years. Forward estimates reflect Bloomberg consensus as of 8 July 2026.
Sources: Freedom Broker (Saken Ismailov); Needham (Laura Martin); Phillip Securities (Yi Qi Lim); Bloomberg Intelligence (Robert Schiffman); Wells Fargo (Alec Chapman); Bloomberg News; Dow Jones Wall Street Journal; Ameriprise Advisor Services (Andrew Heaney); Benzinga; Bloomberg Law; Bloomberg First Word.