Invest1 publisher2 min readPublished
Eight months of investment-grade issuance already fills 88% of the full-year forecast
Investment-grade issuers priced $1.68tn through August, 27% ahead of last year and roughly 88% of the full-year forecast, at coupons that were under 3% for many of the same borrowers a few years ago.
The Investor · Invest desk
What happened
- US investment-grade corporate bond issuance reached roughly $1.68tn through August 2026, running 27% ahead of the same period last year, according to cryptobriefing.com.
- Ten-year Treasury yields hit about 4.94% to 4.97% in early-to-mid September, the highest since 2023, which the article puts down to inflation concerns, an undecided Fed and large fiscal deficits.
- Yields on high-grade corporate debt passed 5.5% at points during 2026, the rate paid by borrowers with the cleanest balance sheets.
- Technology companies funding data centers, specialized chips and AI computing capacity are now the single largest source of new corporate bond supply.
- Credit spreads have narrowed even as benchmark yields climbed, so investors are charging less risk premium on a higher all-in cost of borrowing.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- cost Refinancing the 2021 vintage falls on whoever has that paper maturing, and the higher interest bill is contractual, so it is committed before any board votes on a capital budget.
- contradiction The same article that puts high-grade borrowing above 5.5% names AI infrastructure builders as the biggest source of new supply, so at the largest spenders a dearer cost of capital is showing up as more borrowing.
- exposure If corporate supply outruns Treasury issuance as the forecasters cited by cryptobriefing.com expect, issuers late in the queue bid for the same investor dollars as the government and pay for the crowding.
- decision Treasurers who deferred sales earlier in the year have a mixed record to argue from, so waiting for cheaper money is now a call that has to be defended.
Eight months at $1.68tn is about $210bn a month [13]. Holding the year to just over $1.9tn means September through December prices roughly $220bn between them, about $55bn a month, a quarter of the pace so far [14]. Cryptobriefing.com does not name the forecasters behind the full-year figure, and it gives the $1.68tn as a single number covering refinancing and new issuance together [22].
The 27% increase implies about $1.32tn over the same eight months of 2025, so roughly $357bn of extra paper priced this year [16]. If the largest slice of that is technology companies borrowing for data centers, chips and computing capacity [7], the total says less about treasurers timing rates than the growth rate suggests, because a data center financed at 5.5% is new spending, approved at the higher cost of money [4].
Oracle sold $25bn and took more than $129bn of orders [8], about 5.2 dollars bid for every dollar priced [17]. Set the September 10-year against high-grade yields above 5.5% and the gap is roughly 55 basis points, on figures the article gives for different dates [18]. At that level the paper competes with equity return expectations and sits senior in the capital structure [12].
An issuer rolling a five-year 2021 bond at 3% into this market pays nearly double, cryptobriefing.com says [11]. On $1bn of maturing paper, annual interest goes from $30m to $55m [19].
Some companies postponed sales earlier in 2026 when yields surged, some got the pullback they wanted, and others watched rates climb and came anyway [6]. I think the $1.68tn is a maturity calendar plus AI capex, and not a collective judgement that a 5% 10-year is the ceiling [3]. The counter-case is straightforward: with high-grade money above 5.5% [4], a treasurer who expects worse should issue long now, and a book bid 5.2 times over says the buyers will take it [17]. A tenor and sector split would settle which it is, because a real term-out shows up as long-dated paper from issuers with nothing maturing soon.
Spreads are the test. If they widen from here while issuance holds near $200bn a month, then this year's supply was absorbed at a risk premium that was too thin, and the narrowing was the temporary part [9] [13].
What to watch
- A tenor and sector split of the $1.68tn, which would show whether long-dated non-tech paper is behind the 27%.
- Revisions to the above-$1.9tn full-year forecast as September-to-December supply actually prices.
- Whether corporate bond sales do exceed Treasury issuance, as the forecasters cited by cryptobriefing.com expect.