US investment-grade corporate bond sales hit $145.2 billion in August 2026, a monthly record that eclipses 2020’s pandemic-era peak of $136 billion. It is the third consecutive month in which the market has set a new supply record, according to Bloomberg data. What is driving the deluge is not banks refinancing or utilities funding pipelines. It is a handful of hyperscalers borrowing to build data centers for artificial intelligence.
The number, in context
August is typically a light supply month for high-grade credit. Underwriting desks are half-staffed, portfolio managers are on the beach, and issuers wait for the post-Labor Day window. In 2026 the calendar barely mattered — August alone topped what the entire fourth quarter of some earlier years produced. Full-year 2026 gross investment-grade issuance is now on track to land between $1.8 trillion and $2.25 trillion, with AI-related supply alone estimated at roughly $300 billion, per Wall Street syndicate desks summarized by Morningstar.
The share shift is stark. Between 2022 and 2024, hyperscalers accounted for roughly 2% of USD investment-grade issuance. In 2026 that share is on pace for about 9% — a more than four-fold jump in three years, according to J.P. Morgan Asset Management. Year-to-date, hyperscalers have raised about $219 billion in USD-denominated IG bonds and another $62 billion equivalent in euros, sterling, yen and other currencies.
Alphabet’s $25B was the marquee
Alphabet returned to the debt market in early August with a jumbo deal that sized up to $25 billion across as many as ten tranches, ranging from two years to 40 years. Peak demand hit roughly $115 billion — the third-largest order book of the year, behind Oracle’s February deal ($129B in peak demand) and Amazon’s March sale (about $126B), per Bloomberg. Initial price guidance on the 40-year tranche came at roughly 1.55 percentage points over comparable Treasuries.
The demand story is important. Even as issuance keeps setting records, investment-grade order books remain oversubscribed by three to five times on the marquee deals. The catch is that spreads on existing bonds from the same issuers have to widen to make room for new supply. When Amazon printed its jumbo mid-July tranche, existing Amazon 30-year paper cheapened by about 20 basis points in the secondary market before stabilizing, according to Sage Advisory.
The 2026 mega-deal ledger
| Issuer | Month | Deal size | Peak order book |
|---|---|---|---|
| Oracle | Feb 2026 | $18B+ | ~$129B |
| Amazon | Mar 2026 | ~$54B (multi-tranche) | ~$126B |
| Amazon | Jul 2026 | $25B | Oversubscribed |
| Alphabet | Aug 2026 | $25B (up to 10 tranches) | ~$115B |
| Meta, Microsoft, others | 2026 YTD | Multiple $10B+ deals | 3–5x covered |
Sage Advisory counts nine investment-grade deals larger than $20 billion in 2026 so far — seven of them from hyperscalers. In aggregate, Microsoft, Meta, Google, Oracle, Amazon and Nvidia have more than doubled their combined US-dollar debt footprint to over $360 billion since September 2025.
Why now — and why bonds
The catalyst is cash flow, or the lack of it. Big Tech is on pace to spend more than $730 billion on capital expenditure in 2026, the vast majority earmarked for AI data centers, custom silicon, and the power infrastructure to run them. Alphabet posted its first-ever quarterly negative free cash flow in Q2 2026, per company filings referenced in the Bloomberg coverage of the bond deal.
Even for companies still generating cash, the math favors debt. Coupons around 5% on 10- to 40-year AAA/AA paper are cheap compared to reinvestment yields on cash balances and vastly cheaper than equity dilution at current earnings multiples. When your after-tax cost of debt is a fraction of your marginal return on capital — and AI infrastructure is at least being modeled that way — issuing bonds is the textbook answer.
The spread story is more nuanced than the yield story
Treasury yields have grabbed the headlines. The 10-year note was near 4.67% at the end of August, and the 30-year around 5.19%. Rising term premium — the extra yield investors demand for locking up money at the long end — is doing much of the work, and the tsunami of high-grade supply is one of the reasons the term premium has drifted higher.
Credit spreads themselves, however, remain tight. The ICE BofA US Corporate Index option-adjusted spread printed at 79 basis points in July 2026 — well below the 20-year average — and the BBB subset was around 97 bps in August. In other words, the market is not currently demanding much extra compensation for corporate credit risk despite the supply surge. That could reflect either genuine confidence in hyperscaler cash flows or complacency built on strong recent returns. Which it is depends on how the AI capex cycle plays out.
What to watch next
Three things are worth tracking as the supply pipeline continues into the fall:
- Concessions on jumbo deals. If new-issue concessions to the secondary curve keep widening beyond 15–20 bps, that is the market signaling saturation.
- Ratings agency commentary. Moody’s, S&P and Fitch have so far kept the hyperscalers at top-tier IG. A move to negative outlook on any of the top six would reprice a lot of paper.
- Non-USD share. Hyperscalers issuing more in euros, yen and sterling ($62B YTD) may be an early sign that USD demand is filling up faster than USD issuance capacity.
Sources
- Bloomberg: US High-Grade Bond Sales Set August Record in Year Full of Them (Aug 17, 2026)
- Bloomberg: Alphabet Raises $25 Billion From Sought-After Bond Sale (Aug 6, 2026)
- J.P. Morgan Asset Management: Hyperscalers — Now also a credit story
- Sage Advisory: AI Buildout Fuels Hyperscaler Bond Issuance and Credit Spread Volatility
- Morningstar: Bond Issuance Backing AI Investment Tops $250 Billion
- FRED: ICE BofA US Corporate Index Option-Adjusted Spread
- FRED: ICE BofA BBB US Corporate Index Option-Adjusted Spread
- SIFMA: US Corporate Bonds Statistics
Disclosure: This article is for informational purposes only and is not investment advice.