“Carnage in hyperscaler bond land”, claims the Tweet below. The graph below the Tweet shows “exploding” yield spreads for hyperscalers. These hyperbolic messages and many others are generating fear in some investors that the largest hyperscalers are in financial trouble.
The bearish hyperscaler narrative is not all it’s cracked up to be, as we will explain.
The graph above shows that a basket of credit default swaps** (CDS) for the five largest US hyperscalers is “exploding” from 115 bps to 162 bps in the last few months. Given the basket is comprised of some of the largest US companies (Amazon (AMZN), Meta (META), Microsoft (MSFT), Google (GOOGL), and Oracle (ORCL)), the prospect of a credit event for those companies and it potentially spreading to a broader swath of companies would be troubling.
**CDS is a derivative swap agreement that serves as insurance for bondholders, protecting against a bond default. The graph shows that the cost of insurance for a basket containing Amazon, Meta, Microsoft, Google, and Oracle has risen by almost 50%. As a result, the odds of a default in any of the five names over the next five years have risen from 8.6% to 11.8%.
Background
Before assessing the bearish “carnage” narrative, it’s important to ask why the fear of financial difficulties is increasing. The primary answer is the massive funding needs the hyperscalers require to build out data centers and the poor visibility into the timing and amount of future revenue from those investments.
The amount of capital needed by Amazon, Meta, Google, Microsoft, and Oracle is rising rapidly. The need to raise both debt and equity capital is relatively new because the companies are shifting from cash flow-positive to cash flow-negative positions. Their profits are no longer large enough to cover their AI-related investments.
Currently, the five hyperscalers have combined 2026 capital budgets of roughly $725- $770 billion, up from less than $100 billion in 2021. Alliance Bernstein (AB) estimates that spending from the group could reach nearly $1.6 trillion by 2030.
The graph below shows the sharp decline in free cash flow for the hyperscalers and, equally abrupt, the commensurate surge higher from semiconductor companies that are on the receiving end of their capital expenditures.
Is The Bond Market Concerned?
Growing worries about rising debt issuance are reflected in higher CDS spreads, as we discussed above. While CDS spreads are an important gauge of creditworthiness, it’s important to recognize that CDS trading can be illiquid and volatile, which causes unjustified comfort or concern.
A more robust way to track credit conditions for a company or the broader market is via credit spreads. Credit spreads are the difference between a bond yield or a corporate index yield and a comparable-maturity Treasury bond. A rising spread indicates investors are demanding an additional yield premium to compensate for increased default risk.
The two charts below help us assess whether “carnage” is the appropriate word to describe the bond market’s trading of hyperscaler debt.
The first graph shows the yields of the five hyperscalers alongside US Treasury yields. The second graph answers our question: “Is the bond market concerned?” It charts the hyperscalers’ credit spreads as compared to the credit spread of the corporate bond BOA ICE AA-rated Index.
Excluding Oracle, the four other hyperscalers’ credit spreads have been relatively flat and, since the start of 2025, have been at or below the index representing all AA-rated bonds.
Based on the data, the fear being spread is an Oracle story, not an indictment on Google, Microsoft, Meta, or Amazon. Moreover, even focusing solely on Oracle, “carnage” still overstates the problem in our opinion.
Oracle
As we share below, Oracle’s five-year CDS spread has been rising quickly from below 50 bps in mid-2025 to 200 bps today.
On the cash bond side, an Oracle bond maturing in February 2031 trades at a yield of 6.23%, about 180 bps above the five-year Treasury. Currently, Oracle has a BBB- rating from S&P Global. Based on the table below, courtesy of SimpleVisor (the graphic is from our soon-to-be-released version), Oracle yields are much higher than BBB-rated bonds and in line with junk-rated BB bonds.
Oracle’s 2026 capital expenditure was $55.7 billion. The company has negative free cash flow of $23.7 billion and total debt near $130 billion. Its debt-to-equity ratio is near 4x. As shown below, its key debt ratio is well above the other hyperscalers:
Microsoft .30x
Meta .36x
Google .18x
Amazon .51x
Amazon, Microsoft, Alphabet, and Meta
Unlike Oracle, credit ratings for this group remain firmly in the high investment-grade tier:
Microsoft is AAA (S&P) and Aaa (Moody’s), one of only two U.S. public companies at the AAA rung.
Amazon is AA (S&P), AA- (Fitch), and A1 (Moody’s).
Google is rated AA+ (S&P) and Aa2 (Moody’s).
Meta was recently upgraded to Aa3 (Moody’s) and holds AA- at S&P.
The high credit ratings are not surprising given the low debt-to-equity ratios we shared earlier. According to Credit Sights data, the ratios are well below the approximate .80x average for S&P 500 companies in aggregate.
The credit profile of the four hyperscalers is not a concern today, and further, the word “carnage” does a gross injustice given they are trading better than their corporate bond index. However, we should not get complacent. Rapidly rising debt without commensurate income could change their credit standings rapidly.
Summary
“Carnage” and “Exploding” are dangerous descriptors of what is happening to hyperscalers’ credit spreads. Yes, the basket of CDS spreads is certainly moving upward, but it is not evenly distributed. Oracle’s CDS spreads have moved by multiples, while the other four hyperscalers’ CDS spreads have increased but by a much more subdued amount, and from unusually tight levels to begin with. At face value, the graph greatly overstates the risks facing Microsoft, Amazon, Alphabet, and Meta bondholders.
What the graph really highlights is an Oracle problem. Driven by a balance sheet stretched by debt-fueled AI infrastructure spending and a credit rating teetering a notch above junk, investors are rightfully concerned. To wit, they are pricing Oracle bonds as if they have already been downgraded into junk territory.
The question investors should be asking isn’t whether Oracle is an outlier. It clearly is. The follow-up is whether Oracle is a preview of what happens to credit markets more broadly if AI capital spending keeps outrunning AI revenue.
Comments
Log in or sign up to join the conversation.