Amazon stock falls as $25B bond sale signals AI debt fatigue

Amazon's stock fell approximately 1.1% in Wednesday's regular-session trading as the broader market reassessed the health of hyperscaler debt markets. The catalyst was not Amazon alone but a broader trend: Amazon, Alphabet, Meta Platforms, and Oracle collectively issued approximately $194 billion（約31兆円） of bonds through July 7, according to Reuters' analysis of LSEG data—a 79% increase over the roughly $108 billion（約17兆円） issued during all of 2025.

The underlying concern is one of demand: cover ratios, which measure orders relative to bonds offered, have compressed dramatically. In February, these ratios stood at nearly five times; by July, they had fallen to less than two times. Amazon's own issuance history underscores this deterioration. Its March bond sale was approximately 3.4 times oversubscribed, meaning investors' demand significantly exceeded the bonds available. In stark contrast, its July $25 billion（約4兆円） offering attracted orders equal to only about 1.6 times the amount sold. According to Sage Advisory, Amazon's latest $25 billion（約4兆円） sale widened the spread on one of its earlier 30-year bonds by approximately 20 basis points—a modest but visible repricing upward.

The weakness extends across the entire hyperscaler universe. Among 91 comparable hyperscaler bonds issued during 2026, 78 are trading at higher yields than at their issuance price, with a median increase of approximately 22 basis points. This pattern reflects investor unease about the sector's debt trajectory. Goldman Sachs estimates that debt currently funds roughly one-third of hyperscaler capital expenditure, underlining the sector's reliance on market access. Looking ahead, Goldman projects that these companies and Microsoft will issue approximately $250 billion（約40兆円） during 2026 and $400 billion（約64兆円） in 2027—levels that assume continued capital-market access at manageable costs.

Amazon will report second-quarter results on Thursday, July 30, at 5 p.m. ET. Investors will scrutinize whether cloud growth and operating cash generation can justify the substantial borrowing the company has committed to for its AI infrastructure plans. If cash generation disappoints or cloud momentum slows, the market's willingness to fund hyperscaler AI capex at current terms may weaken further.

Hyperscalers' ability to access debt markets at favorable terms has eroded sharply over recent months. Cover ratios—a key metric tracking investor demand—have fallen from nearly five times in February to less than two times in July, reflecting a dramatic shift in market sentiment. Amazon's own bond experience illustrates this trend: its March offering was approximately 3.4 times oversubscribed, meaning demand far exceeded supply, while July's $25 billion（約4兆円） sale drew orders for only about 1.6 times the bonds offered. This weakening demand is not isolated to Amazon; among 91 comparable hyperscaler bonds issued during 2026, 78 are trading at higher yields than at issuance, with a median increase of approximately 22 basis points, indicating that investors are repricing the risk of these borrowings upward.

The shift matters because debt is now funding roughly one-third of current hyperscaler capital expenditure according to Goldman Sachs estimates. As these companies pivot to massive AI infrastructure investments, they depend on favorable borrowing conditions to support their spending plans. Goldman projects that hyperscalers and Microsoft will together issue approximately $250 billion（約40兆円） in bonds during 2026 and $400 billion（約64兆円） in 2027—a substantial increase from recent issuance levels. If investor demand continues to deteriorate or cover ratios remain compressed, these companies may face higher borrowing costs or difficulty accessing the capital markets at the scale they require.

No comments yet. Be the first to share your thoughts!