NextFin News - US investment-grade bond sales have posted a third straight monthly record in August 2026, with $1.4 trillion of notes sold year-to-date — 9% ahead of the pace set in 2020, the year that ended at a record $1.75 trillion. Global publicly syndicated issuance is on track to hit $5 trillion more than a month faster than the prior best set last year, and Milwood Hobbs Jr., Oaktree Capital Management's deputy chief investment officer of strategic credit, says the force behind the surge is not cheap money but momentum itself.
The easy narrative would be that an accommodative Federal Reserve unlocked a refinancing wave. The reality is more complicated: the Fed cut rates three times in late 2025, then held its benchmark rate at 3.5%-3.75% through July 2026 — where the vote ran 9-3, with three regional presidents preferring a quarter-point hike. Borrowing costs are not collapsing; demand is simply overwhelming supply. That distinction is what makes Hobbs's read of the market worth weighing. In his view, spreads have found a floor, liquidity is abundant, and the market is moving because it is moving. The question for investors is whether that momentum is a cyclical wave that will crest once AI capital spending slows, or a structural shift in how the world's largest companies fund themselves.
The Numbers Behind the Record
US investment-grade issuance opened 2026 with $208.4 billion in January, a record for the month and a 12% increase over the prior January, and each subsequent month has built on that start. Through July, total US corporate bond issuance reached $1.681 trillion, up 26.9% from the same period a year earlier, according to data from the Securities Industry and Financial Markets Association. The August rush defied the calendar: the week of August 10 alone saw $80 billion of supply, the third-highest week of 2026, in a month that since 2019 has averaged just $95 billion for the entire period. On that Monday, nineteen firms priced investment-grade bonds — the most single-day borrowers in seven months, surpassed only by the twenty deals that hit the market on January 5 in a typical start-of-year rush.
The composition of the borrowing is what separates this cycle from the last refinancing wave. Hyperscale technology companies — Microsoft, Meta, Alphabet, Amazon and Nvidia — have collectively issued roughly $225 billion in bonds year-to-date as of mid-2026, a 973.7% increase from the prior year. Microsoft set the tone on August 1 with a $19.75 billion multi-tranche investment-grade offering, the fifth-largest corporate bond sale in global history. SpaceX raised $25 billion in a single June issuance. Full-year hyperscaler issuance is projected to approach $400 billion, up from $121 billion across all of 2025.
Spreads now have a floor, creating opportunities.
— Milwood Hobbs Jr., Deputy Chief Investment Officer of Oaktree's Strategic Credit platform, speaking at PEI's NEXUS 2026 conference.
Wall Street's forecasts have been revised upward alongside the tape. JPMorgan projects a record $1.81 trillion of US investment-grade issuance in 2026, a 17% increase. Morgan Stanley strategists see supply exceeding $2 trillion, with AI-related debt more than doubling to $400 billion. M&A-linked issuance is forecast at $182 billion, up 21% year over year, as deals including AT&T's acquisition of Lumen Technologies' consumer fiber operations and Kraft Heinz's planned grocery spinoff move forward.
Why Momentum, Not Just Low Rates, Is the Driver
The transmission mechanism runs through three channels, and the first is scale. The AI buildout has converted corporate bond markets into the primary funding pipe for a capital cycle that market research firms estimate could reach $3 trillion across AI, cloud computing and data-center infrastructure by 2029. Debt is the preferred tool because equity issuance at current valuations is dilutive, and private credit alone cannot absorb deals of this size. A $20 billion Microsoft offering or a $25 billion SpaceX sale simply cannot be placed in the middle market; the syndicated investment-grade market is the only venue with the depth. That concentration of volume in the largest names is what turns a steady flow of issuance into a record.
The second channel is technical and self-reinforcing. Strong order books on new issues mean companies do not need to offer meaningful concessions, which keeps spreads tight, which draws in more demand, which keeps issuance cheap. Hobbs's point that spreads "have a floor" captures the asymmetry: investors, sitting on cash and facing a central bank that has already eased, are reluctant to bet against credit, so every dip in prices is bought. That is momentum in its purest form — the market rising because it has risen, with each successful deal validating the next. It is also why the Fed's rate path matters less than the flow path: as long as deals clear, the incentive to issue remains.
The third channel is a shift in funding behavior that outlasts any single rate cycle. Companies are locking in longer-dated debt now not merely to refinance but to finance multi-year capital programs whose payoffs lie years ahead. That changes the duration and purpose of the liability stack. A refinancing wave is cyclical by nature — once the old debt is rolled, the urgency fades. A capital-expenditure financing wave tied to a technology transition is different: the borrowing continues as long as the buildout continues, and the buildout is measured in years, not quarters.
The Cyclical Case: This Is a Wave That Will Crest
The strongest argument against treating this as structural is that the borrowing is tied to a single, concentrated theme. Hyperscaler issuance is up nearly tenfold in a year. When a single driver accounts for that much of the volume, the cycle is only as durable as the driver. If AI capital spending growth slows — because model-training costs fall, because utilization lags the buildout, or because regulators constrain data-center power — the issuance pipeline thins quickly. A market that set records on $225 billion of hyperscaler supply has more to lose from a slowdown than a market diversified across sectors.
There is also the question of demand sustainability. Bank of America analysts Yuri Seliger and Sohyun Marie Lee warned that credit quality, which strengthened after companies slashed leverage on recession fears, is "poised to start eroding" as companies pursue heavy-debt-financed deals. "We expect weaker fundamentals and technicals in 2026," they wrote. "That means spreads should trade a bit wider." TD Securities' Hans Mikkelsen forecasts spreads widening to 1.0-1.1 percentage points in 2026 from 0.75-0.85 percentage points, as the "period of excess demand ends." Deutsche Bank sees European spreads moving to 1.03 percentage points by year-end from 0.86 percentage points.
History offers a cautionary parallel. The 2020-2021 issuance boom, also driven by cheap rates and a technology capital-expenditure narrative, gave way to a sharp slowdown once inflation forced central banks to reverse course. The difference this time is that the Fed is not hiking — but the three dissenters at the July meeting, who preferred a quarter-point increase, are a reminder that the policy direction is contested. The lesson that a concentrated borrowing theme can reverse quickly still applies.
The Structural Case: Funding Has Changed for Good
Hobbs's own record points the other way. Before joining Oaktree in 2013, he built the high-yield sales and trading franchise at Natixis and worked in leveraged finance origination at Goldman Sachs and Deutsche Bank. His view that private credit has "an abundance of liquidity" and that volatility could ignite M&A rests on a deeper shift: private equity sponsors hold roughly $4 trillion of dry powder, and private credit has grown into a permanent funding layer alongside the syndicated market. That is not a cyclical condition; it is a change in market structure.
The maturity wall ahead reinforces the structural read. Around $200 billion of private credit loans were written in 2021 and 2022 at peak tech valuations, when interest rates were near zero. Those loans mature in 2027 and 2028, after rates moved from zero to 5%. Hobbs has pointed to that cohort as the place to "focus on potential dislocation." When those borrowers refinance, many will need to tap the syndicated investment-grade or high-yield markets rather than private credit, adding a multi-year source of supply that is independent of the AI cycle. That wall exists regardless of whether hyperscaler capex grows, slows, or stalls.
In other words, even if the AI borrowing wave proves cyclical, the market has added a structural source of volume: a larger, more liquid private credit complex feeding into public markets, and a maturity wall that will force refinancing regardless of the rate path. The two forces can coexist — a cyclical AI leg riding on top of a structural rise in the issuance baseline.
Second-Order Effects: What the Market Has Not Priced
The first-order read — cheap debt funds AI — is already in the tape. The second-order question is what happens to the rest of the credit complex when the hyperscalers dominate investor attention and allocation. When $225 billion of the cleanest, most liquid paper is available from the world's strongest balance sheets, capital flows toward it. That can starve lower-rated borrowers of demand even as the headline investment-grade market breaks records. The record is not broad-based strength; it is concentrated strength, and concentration is fragile.
The third-order effect runs through the central bank. If the Fed is holding steady while a credit boom extends, it is arguably tolerating leverage at a late stage of the cycle. The market is treating the 2025 cuts as a tailwind for issuance; the risk is that policymakers who wanted to hike in July are seeing something the credit market has not yet priced. A preventive cut supports asset prices; a reactive pause in the face of rising leverage does not. The distinction matters more for credit than for equities, because credit returns are capped on the upside and exposed on the downside.
What Would Prove the Thesis Wrong
The momentum thesis fails if issuance momentum breaks before the AI cycle does. A specific falsifying signal: if US investment-grade monthly issuance falls below $100 billion for two consecutive months while hyperscaler capital-expenditure guidance remains unchanged, the "momentum" explanation collapses and the cycle is being driven by fundamentals, not self-reinforcing flows. A second signal: if investment-grade spreads widen beyond 1.1 percentage points before any deterioration in default rates, the technical bid has already broken and the floor Hobbs describes has given way.
Outlook: Three Horizons
Short term (next quarter): Momentum favors continued heavy issuance. The post-Labor Day window typically brings a fresh burst of supply, and order books remain well-covered. Base case: another strong quarter, with full-year US investment-grade issuance landing near JPMorgan's $1.81 trillion forecast. Upside case: issuance exceeds $2 trillion if M&A activity accelerates alongside AI spending. Downside case: a market dislocation in equities or a Treasury auction failure slows syndication.
Medium term (2027-2028): The private credit maturity wall arrives. Refinancing demand from the 2021-2022 vintage adds supply even if AI capex moderates. Upside case: refinancing is absorbed smoothly as rates stabilize and spreads remain contained. Downside case: a weaker growth backdrop meets the maturity wall, and Hobbs's "potential dislocation" materializes in spreads rather than defaults.
Long term (structural): The funding mix has changed. Private credit is now a permanent, liquid layer of the capital structure, and large technology capital programs are funded with bond markets as a matter of course. That supports a higher baseline of issuance than the pre-2020 era, even after the AI wave normalizes.
For investors, the asymmetry is clear: the beneficiaries are the arrangers and investors who can place and hold large, liquid investment-grade paper; the exposed are the holders of lower-rated credit who may find demand migrating up the quality ladder. For borrowers, the window is open — but windows that stay open on momentum alone tend to close quickly when the momentum stops.
The record itself is not the story. The story is that the market is being carried by a funding cycle tied to a single technology bet, and momentum is a fair-weather friend. Hobbs is right that spreads have a floor today; the question is whether that floor is built on durable structural demand or on the temporary absence of sellers. If hyperscaler issuance slows while the Fed holds rates steady, the answer will become obvious — and the momentum trade will have nowhere to run.
Explore more exclusive insights at nextfin.ai.
