---
title: "AI Bond Issuance Wave Continues, but Market Becomes \"More Selective\""
type: "News"
locale: "en"
url: "https://longbridge.com/en/news/293988435.md"
description: "UBS Group AG believes that the credit market's capacity to absorb new bond issuances is declining significantly. Amazon's latest investment-grade bond issuance in July saw a subscription ratio of only 1.6x, far below the 3.4x recorded in March. Spreads on several long-dated technology bonds widened by 20 to 50 basis points after secondary market trading. UBS expects long-end spreads to widen by another 10 to 15 basis points over the next six months, as the market shifts from an \"accept-all\" stance to being \"highly selective.\""
datetime: "2026-07-28T00:44:37.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/293988435.md)
  - [en](https://longbridge.com/en/news/293988435.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/293988435.md)
---

# AI Bond Issuance Wave Continues, but Market Becomes "More Selective"

The AI-driven bond issuance boom has not ended, but cracks are beginning to appear.

According to Zhuifeng Trading Desk, UBS Group AG pointed out in its global strategy report on July 27 that although capital expenditure forecasts for hyperscalers continue to be revised upward—totaling approximately $849 billion in 2026 and potentially exceeding $1 trillion in 2027—**the credit market's capacity to absorb new bond issuances is declining significantly, with massive new supply facing challenges from weakening demand and poor secondary market performance.**

UBS believes that **the market is repricing the "long-term monetization risk" of AI investments**. **On the demand side**, Amazon's latest investment-grade bond issuance in July had a subscription ratio of only 1.6x, far below the 3.4x in March; spreads on several long-dated technology bonds widened by 20 to 50 basis points after entering the secondary market. **In the secondary market**, the credit curve for investment-grade (IG) technology bonds is steepening sharply, requiring larger discounts to attract capital for new issuances.

## Issuance Forecasts Revised Upward Again: AI Capex "Only Increases, Never Decreases"

As Alphabet, Microsoft, Meta, and Amazon successively announced their financial results, capital expenditure guidance continued to exceed expectations, becoming the core focus of the credit market.

In its early 2026 outlook, UBS had already **raised its forecast for total US investment-grade technology bond issuance to $450 billion**, reflecting the expanding range of issuers participating in the AI capital expenditure cycle.

Furthermore, as Alphabet raised its fiscal 2026 capital expenditure guidance to $195–205 billion, and Microsoft increased its fiscal 2027 capital expenditure forecast from $234 billion to $261 billion, **UBS further raised its forecast for debt issuance contributions from hyperscalers to $275–285 billion (previously $230–240 billion).**

The combined fiscal 2026 capital expenditure forecast for the six major hyperscalers (Alphabet, Amazon, Microsoft, Meta, Oracle, CoreWeave) has reached $849 billion.

> Among them, Google and Amazon have dominated issuance year-to-date, while Microsoft and Meta are expected to accelerate their issuance pace in the second half of the year. Oracle is an exception—its leverage ratio has exceeded 4x, and it was recently downgraded to BBB- by S&P. **UBS expects no new bond issuances from Oracle this year, although off-balance-sheet financing arrangements cannot be ruled out.**

Notably, the proportion of technology bond issuances this year with maturities exceeding 10 years reached 32%, significantly higher than the historical average of 19% since 2024. This indicates that issuers generally view AI-related capital expenditures as a multi-year cycle and tend to extend debt duration.

## Clear Signals of Cooling Demand: Market Begins to "Vote with Its Feet"

Despite the continuous expansion in issuance scale, the credit market's willingness to absorb new technology bonds has shown significant divergence.

The most intuitive data comes from Amazon: **its investment-grade bond issuance in July had a subscription ratio of only 1.6x, compared to a high of 3.4x for the same issuer in March this year.**

UBS attributes this phenomenon to three factors:

> -   **First, fundamental pressures**. Corporate-side "Token optimization" (i.e., users shifting to lower-cost AI models), memory/hardware inflation, and migration to cheaper alternatives may suppress the profitability of AI frontier labs and hyperscalers in the near term.
>     
> -   **Second, uncertainty in monetization timelines**. As the scale of debt issuance continues to climb, investor doubts are deepening regarding whether recent massive capital expenditures can be converted into substantial returns within a reasonable timeframe, leading investors to demand higher long-term risk premiums.
>     
> -   **Third, the amplifying effect of interest rate volatility**. The yield on 30-year US Treasuries rose by approximately 30 basis points over the past month, further amplifying price pressure on long-duration investment-grade bonds.
>     

Meanwhile, the recent secondary market performance of several technology bonds is concerning, especially for long-dated instruments with maturities of over 20 years. Spreads on some bonds widened by 20 to 50 basis points after issuance, but UBS believes that **the adjustment is not yet complete.**

Currently, the average steepness of the 10-year/30-year credit curve for hyperscalers is about **42 basis points**, higher than the overall average of **approximately 22 basis points** for multi-maturity investment-grade issuers this year.

However, historically speaking, this level is far from extreme:

> During the M&A boom from 2013 to 2015, the steepness of the investment-grade 10-year/30-year curve reached **80 to 85 basis points**; it also reached **60 to 65 basis points** during the TLAC regulatory reforms for financial institutions in late 2016.
> 
> 
UBS expects that, against the backdrop of capital expenditures continuing to exceed expectations, negative free cash flow, and uncertain investment return timelines, long-end bond spreads for hyperscalers have room to widen by another 10 to 15 basis points over the next six months or so.

From the CDS market perspective, the credit market has begun to show significant divergence since March: **Google and Amazon's CDS spreads are 10 to 15 basis points wider than Microsoft's**, while Meta's CDS spreads are about 20 basis points wider than the former due to relatively insufficient observable evidence of capital expenditure monetization.

## Four Chain Effects: Spreading from Technology Bonds to the Entire Long-End IG Market

UBS pointed out four potential impacts of the recent poor performance of technology bonds on the entire investment-grade market in its report:

> **First, new bond issuances require larger issuance concessions**.
> 
> Given that hyperscalers still have substantial financing needs and market concentration is increasing, issuers may need to offer larger discounts to attract demand. The "first-mover advantage" effect may become more pronounced—early issuers are more likely to lock in subscriptions, while latecomers face the risk of market crowding.
> 
> **Second, the credit market will reward issuers that can demonstrate evidence of capital expenditure monetization**.
> 
> During this earnings season, investors will focus on three key indicators: **sustained growth in cloud revenue, the stickiness and growth of RPO (Remaining Performance Obligations)/backlog, and whether margins are being compressed.**
> 
> **Third, the impact weight of AI-related bonds on the overall performance of investment-grade indices continues to rise**.
> 
> Currently, the market cap weight of the five major investment-grade hyperscalers in investment-grade indices is about 4.5%, but the duration-adjusted DV01 weight has reached 6% (2.6% and 3.3% respectively in 2024), comparable to the four major money center banks. If this year's issuance forecasts are realized, this proportion could rise to 7% to 8% by year-end.
> 
> **Fourth, the expansion of long-end technology bond supply may trigger rotational selling in other sectors**.
> 
> In investment-grade indices, for bonds with maturities of over 20 years, consumer staples account for about 19%, utilities and communications each about 14%, energy about 8%, and capital goods about 5%, excluding technology. UBS expects that **hyperscalers issuing long-term bonds at wider spreads will first trigger rotational selling in the technology and communications sectors, subsequently affecting AI-sensitive sectors such as utilities and capital goods.**
> 
> 
In summary, UBS believes that **the AI debt financing cycle has not ended; this is merely a "speed bump," not the end**. However, the market has **shifted from an "accept-all" stance to being "highly selective"**. Investors need to seek targets within the AI industry chain that can convert capital expenditures into cash flows faster and with greater certainty, rather than continuing to bet on the long-end bonds of hyperscalers with highly uncertain monetization timelines.

### Related Stocks

- [AMZN.US](https://longbridge.com/en/quote/AMZN.US.md)
- [UBS.US](https://longbridge.com/en/quote/UBS.US.md)
- [AMZW.US](https://longbridge.com/en/quote/AMZW.US.md)
- [AMZZ.US](https://longbridge.com/en/quote/AMZZ.US.md)
- [AMZD.US](https://longbridge.com/en/quote/AMZD.US.md)
- [AMZO.US](https://longbridge.com/en/quote/AMZO.US.md)
- [AMZU.US](https://longbridge.com/en/quote/AMZU.US.md)
- [AMZY.US](https://longbridge.com/en/quote/AMZY.US.md)
- [AZYY.US](https://longbridge.com/en/quote/AZYY.US.md)
- [AMZP.US](https://longbridge.com/en/quote/AMZP.US.md)
- [MAGX.US](https://longbridge.com/en/quote/MAGX.US.md)
- [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md)
- [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md)
- [MSFT.US](https://longbridge.com/en/quote/MSFT.US.md)
- [META.US](https://longbridge.com/en/quote/META.US.md)
- [ORCL.US](https://longbridge.com/en/quote/ORCL.US.md)
- [CRWV.US](https://longbridge.com/en/quote/CRWV.US.md)
- [ORCL-D.US](https://longbridge.com/en/quote/ORCL-D.US.md)

## Related News & Research

- [Prediction: CEO Andy Jassy will raise Amazon's full-year 2026 capex guide on July 30](https://longbridge.com/en/news/293883787.md)
- [Amazon cracks down on use of AI images by sellers after New York law](https://longbridge.com/en/news/293681872.md)
- [Discover top deals with AI-powered shopping this Great Freedom Sale on Amazon.in starting 7th August 2026](https://longbridge.com/en/news/294095182.md)
- [UBS 2Q FY26 net profit rises to USD 2.8 billion; core businesses’ underlying pretax profit climbs 47% year on year](https://longbridge.com/en/news/294152491.md)
- [UBS lifts Wienerberger voting rights to 5.09% from zero, filing shows](https://longbridge.com/en/news/294203797.md)