$25B Bond Subscription Ratio at Just 1.6x: Is the AI Funding Frenzy Taking a Break? [Analysis]

https://finance.sina.com.cn/stock/hkstock/hkstocknews/2026-07-08/doc-inihckru4748950.shtml

After reading this report on Amazon’s bond issuance meeting a cold reception, there are a few data points worth highlighting separately.

First, look at the key numbers:

• Tech companies have already issued 7 bonds of $25 billion or more this year, exceeding the total of the previous six years combined.
• Amazon’s subscription ratio was only 1.6x, far below the enthusiasm seen in similar financings in March this year, and lower than the average for large-cap tech debt.
• AI-concept companies have raised approximately $182 billion cumulatively this year, which is 14 times the less than $13 billion raised during the same period last year.
• These bonds already account for nearly 15% of total US investment-grade bond issuance in 2026, contributing to more than half of this year’s new supply.

---

A straightforward inference: Supply is accelerating, while demand is showing marginal weakness. Quoting Wellington Management’s view hits the nail on the head—“Investors are unwilling to allocate too high a position to a single issuer because they will likely return to raise funds again in a few months.” This actually reflects the market’s implicit concern about the “unsustainable financing pace” of tech giants.

After SpaceX’s $25 billion bond issuance last month, secondary market performance weakened, with some investors taking profits. This is uncommon among large investment-grade bonds and further validates that “aesthetic fatigue” has appeared on the demand side.

---

But looking at it from another angle, for issuers, entering the bond market only once or twice a year can indeed reduce financing costs and issuance risks. Moreover, new bonds carry a premium of 12-22 basis points over existing bonds (the market average is only 4 bps), which remains attractive to yield-seeking institutional investors. Amundi Asset Management also mentioned that these large-scale issuances have actually improved market liquidity, creating relative value trading opportunities.

A signal worth tracking long-term: Loop Capital noted that as the AI investment cycle matures, the market will pay more attention to project profitability. Currently, most of these AI bonds have just been issued and haven’t yet undergone economic cycles or profit tests. JPMorgan estimates that AI infrastructure investment will require $5.5 trillion by 2030, with $2.1 trillion needing to be completed via investment-grade debt—if this happens at the current pace, it could put greater pressure on the market.

In summary, this isn’t an “AI financing bubble bursting,” but rather a transition from the frenzy phase into a “valuation digestion + supply stress test” stage. Investors need to re-weigh: How much free cash flow can the AI assets behind these bonds generate in the future to cover interest?