In short: Decompose high yield into AI and non-AI: non-AI spreads "have barely widened," but AI junk bonds are "out about 50 basis points from their tights" and AI bank loans "more like 130." Hyperscaler bonds were rejected within days of issue ("75 over, my bid's 200 over"), yet borrowers "won't care if the rates go up 200 basis points."
A "spread" is the extra interest a borrower pays over the government. Gundlach splits the junk-bond and loan markets in two: for non-AI companies the extra interest has hardly moved, but for AI-related borrowers it has jumped (about half a point on bonds, 1.3 points on loans). Some hyperscaler bonds were being bid at far worse prices within days of being sold. That gap is the market quietly flagging where the risk sits.
5:38So when you decompose the triple C sector you start to see things like, just take the high yield sector broadly, forget just the triple C's. If you split into two pieces, AI related borrowing in the bond market and everything in the junk bond market and the bank loan market other than the AI sector, you're starting to see the non-AI sector is still strong.
In short: Duration- and rating-adjusted, non-AI IG spreads went 70 → 78bp "which is almost nothing," while AI IG widened 50bp; ex-AI high yield "really not widened at all," AI high yield +150bp. Spreads blow out "days after a bond gets issued" — "corporate bond investors do not believe the ratings," and it's "not to one notch lower."
A "spread" is the extra interest a company pays over the US government. For non-AI companies it has barely moved. For AI-related borrowers it has jumped — half a point on investment-grade bonds and a point and a half on junk bonds — often within days of the bonds being sold. That tells him buyers don't trust the credit ratings stamped on AI debt and are treating it as much riskier than advertised.
14:25So we've had debt go up by a lot. Of course we all know the Treasury bonds are going up with now bumping against the $41 trillion debt ceiling. There's just a lot of bonds to absorb. And what we're noticing interestingly in the credit market is in the investment grade bond market we divide it into the AI sector of corporate bonds investment grade and then everything but the AI sector and on a duration and rating adjusted basis the spreads on the non-AI sector of investment grade are
In short: "Monumental spread widening": AI-sector IG from ~50 to ~125bp and HY from ~180 to ~325bp while ex-AI spreads are unchanged or near their tights. An avalanche of AI issuance on top of Treasury borrowing — "the market is having a hard time digesting this amount of supply" and is "starting to demand higher compensation"; unclear who's buying (perhaps PE-owned captive insurers).
A credit "spread" is the extra interest a company pays over the government to borrow. For most US companies that premium hasn't budged, but for AI and AI-related borrowers it has more than doubled (investment grade 0.5 → 1.25 percentage points; junk 1.8 → 3.25). AI companies are issuing a flood of bonds to fund data centers, on top of heavy Treasury borrowing, and buyers are demanding more to absorb it. Gundlach flags it as a warning sign to watch closely.
9:07And here we see that on the left hand side of the display the lighter blue line is the spread on investment grade bonds in the US corporate bond market excluding the AI sector. And then we see the darker line which is the AI sector only. And one thing that's fairly clear is that the broader investment grade bond market has not really seen any meaningful spread widening but the AI market has seen monumental spread widening vis-a-vis the investment grade area. We see that the AI spread was only
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.