Bloomberg Surveillance Sep 21, 2026
With Amanda Lynam, Chief Credit Strategist at Goldman Sachs
The hyperscalers are the borrowers everyone is watching, and on Goldman Sachs' own work they account for 40% of the AI-related debt coming through global corporate credit markets. The other 60% is everybody else.
The usual defensive move when a single theme floods a market is to go up in quality. Amanda Lynam is doing the opposite, and she has two specific reasons.
"We've counted nearly 600 billion of global AI-related debt issuance so far this year, and we think it's poised to accelerate even further in 2027."
Lynam is Goldman Sachs' chief credit strategist, and her 2027 issuance forecast is built directly off her own equity analysts' capex estimates with an assumption about how much of it gets debt financed. She expects this to be a feature of the credit market through 2030 and 2031.
The full segment is covered here so you can skip it.
Here are the 6 insights that matter.
Key Takeaways
Hyperscalers are 40% of AI-related credit issuance, not the whole of it — the rest is spread across the global corporate market
Nearly $600B of AI-related debt has been issued globally this year, and she expects 2027 to be bigger
She is overweight BBBs in investment grade and single Bs in high yield, because moving up in quality now means buying the supply and losing the spread cushion
Credit should be bought for yield and income, not for a total-return kick from tighter spreads
The issuance runs through 2030 and 2031, which is why she expects private credit, private infrastructure and private real estate to take a bigger share
What is different this time is that there is no end in sight — the credit market funds re-leveraging best when the need is quantifiable and followed by debt reduction
1. $600B and Counting
The host's framing was that 2026's story is new issuance from the AI players — companies with cash-rich balance sheets that never needed the bond market before.
Why the borrowers changed
A lot of tech companies who you don't normally see in the bond market because you have such tons of cash on their balance sheet, lots of free cash flow. They didn't need the bond market. Now they do for all this AI capex.
Bloomberg Surveillance
Lynam's first move was to correct the scale of what is being watched.
The hyperscalers are the visible part, not the whole
our work suggests that as sizable as those capex needs are from the hyperscalers, they represent just 40 percent of the AI-related issuance that's coming through the global corporate credit markets.
Amanda Lynam
The acceleration started in 2025 and intensified through 2026.
The running total
We've counted nearly 600 billion of global AI-related debt issuance so far this year, and we think it's poised to accelerate even further in 2027.
Amanda Lynam
Her conclusion from that is about what an investor buys instead.
Everything outside the theme is worth more now
What that means is that the diversification value of sectors outside of that theme, so banks, healthcare, energy, food and beverage, it's all the more important for investors.
Amanda Lynam
2. Down in Quality on Purpose
Asked about the gap between AI-related spreads and the rest of the market with the Fed hiking again, Lynam said the bifurcation has changed what being defensive means. She is overweight BBBs against the higher-rated cohorts in investment grade, and overweight single Bs against double Bs in US high yield.
The old reflex now buys you two problems
While the reflex historically to be more defensive or move up in quality was to move up in ratings, if you do that in this environment, what you do is you subject yourself to two more pronounced headwinds.
Amanda Lynam
The first is supply: the higher-rated cohorts of both investment grade and high yield have produced a disproportionately large share of issuance relative to previous years, and she attributes that largely to the AI theme — even in high yield. The second is duration.
Tight spreads leave nothing to absorb a rate move
you lose that spread cushion to buffer total returns from higher rates.
Amanda Lynam
She is careful about the bottom of the market, which she says persistently lags, and says the macro supports the position.
Why she is comfortable there
But growth is good enough in our view. Credit fundamentals are solid.
Amanda Lynam
And what credit is actually for right now
if you're allocating to credit, you should be doing that for yield and income, not for a potential total return boost from tighter spreads or lower rates.
Amanda Lynam
3. The Forecast's Machinery
Paul Sweeney asked what Goldman's tech analysts are telling her about how long the capex cycle lasts, given that her market is being asked to fund it. Lynam described where her own numbers come from.
The issuance forecast is a capex forecast with an assumption on top
our debt issuance forecast for 2027 and even through the outer years, they're directly informed in large part by the capex estimates of our equity team. And then we assume a share of that capex spend that will be debt financed.
Amanda Lynam
And the duration of the theme
this is a multi-year trend. This is something that it's not going to be kind of a one and done wave of issuance. We expect this to be a present theme in the corporate credit markets through 2030, even 2031.
Amanda Lynam
4. Private Credit Steps In
The consequence of a wave that long, on her reading, is that the public corporate bond market cannot carry all of it.
Where the rest of the money has to come from
We think that other markets, in particular other financing channels, will need to play a larger role in financing some of this need, specifically the private credit markets, private infrastructure, private real estate
Amanda Lynam
Her number for the capital sitting in those channels is $4.5 trillion of dry powder across all private market strategies, private equity included. She is not writing off the traditional route: she puts at least another $500 billion of runway in US investment grade alone. The point is that no single channel is big enough.
The shape of the answer
we believe a wide range of financing channels will need to participate across structures, markets, and even currencies.
Amanda Lynam
5. Not a Red Flag, but New
Asked directly whether she sees red flags in AI-related issuance, Lynam declined the word and substituted a comparison.
What the credit market is built to fund
the credit market works best in funding periods of active re-leveraging when the need is quantifiable, when there's an end in sight,
Amanda Lynam
Ideally, she added, a period of debt reduction follows the issuance. The usual example is a jumbo deal funding an acquisition: the deal closes, the issuance stops, and the borrower turns to paying it down.
Why this one does not fit the template
this multi-year period of AI-related debt issuance, where capex estimates have been moving higher, by the way, it's just a different phenomenon than what the corporate credit market is used to.
Amanda Lynam
That, rather than any single credit, is why she expects the private channels to matter more as it extends.
6. Is This the New Normal?
Paul Sweeney's last question was whether AI spending settles back to pre-AI capex levels once the data centers are built. Lynam did not pretend to know.
She declined the question
I think it's hard to know, and it's probably out of my wheelhouse. I'll defer to the tech experts on that one.
Amanda Lynam
What she would answer is the version of it that belongs to her market: whether there is a financing runway for the spend. Her answer is yes, and the reason is that the theme is still small relative to the whole.
The share that surprises her
Tech is actually still a pretty small share of the overall corporate credit market, despite the issuance numbers that I mentioned at the start, which is pretty astounding.
Amanda Lynam
The conclusion she wanted on the record
we are not concerned about access to capital for the AI theme from a corporate credit side.
Amanda Lynam
What she expects instead is a more selective market: investors picking their spots, and more nuanced conversations about where to take the risk and at what price as the cycle runs on.
Bonus Insights
She agreed with the compliment
Told she has some of the best tech research analysts on Global Wall Street, Lynam did not demur.
We sure do have the best equity research tech analysts.
Amanda Lynam
Demand for the paper is not the constraint
Her read from the investor side is that there is a lot of appetite for AI-related bonds. The question she expects to dominate is not whether the capital shows up but which structure, market and currency it shows up in.
Lynam's bottom line is that the AI build-out is a financing event the corporate credit market can absorb, but only by pulling in private credit and other channels — and that the right way to own credit through it is for the income, not for the spread.
Products, Companies & Tools Mentioned
Goldman Sachs (Her firm; its equity analysts' capex estimates are the input to her debt-issuance forecast)
Federal Reserve (The rate backdrop behind her duration argument — tight spreads leave no cushion when rates rise)
Listen to the full episode
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: