Goldman Sachs's Amanda Lynam Talks Private Credit

**SPEAKER_1** (0:02)
Bloomberg Audio Studios, Podcasts, Radio, News.

**Tom Keene** (0:07)
Driving all the credit, Amanda Lynam at Goldman Sachs is in studio with us. It could be a one-hour conversation, we're going to do it in seven minutes here. Your note finally out, it's like a Marty Fridson 101 note. It's like oops, defaults, credit, left tail.
Fold in private credit in the angst over the weekend on private credit.
Public credit, private credit, is there a left tail risk?

**Amanda Lynam** (0:32)
Good morning. Thank you for having me.
So taking them in two. So in the high-yield market, we did raise our default forecast last week because we're a little bit concerned about this left tail of borrowers that haven't been contributing to the overall resilience in the credit markets. That coupled with higher AI related issuance in the high-yield market, that market's not immune, higher commodity costs, higher rates, translating into a higher cost of capital, leaves us on the margin somewhat concerned. On the private credit point, in many ways, we actually just treat this like broader credit. We have flagged recently that non-accruals in private credit have increased a little bit in the first quarter, but it's not outsized relative to the broader trend. The key point we are watching in private credit and the leverage loan market is the 2020 maturity wall because there's a lot of software debt that needs to be refinanced. So far, that refinancing has been encouraging.

**Tom Keene** (1:27)
You can keep your job for 18 months.

**Amanda Lynam** (1:30)
We've already started chipping away at that. It's been encouraging, but that's the key point we're watching.

**Paul Sweeney** (1:34)
All right, Tom, and Amanda's latest report, Exhibit 5
We estimate nearly $200 billion of data center deal activity in the private market since the start of 2025 Man, I did not know that. Who's buying this stuff?

**Amanda Lynam** (1:46)
It's happening under the surface above and beyond the very meaningful supply that we've already had from the AI ecosystem. We estimate that's $500 billion year to date. So just the numbers here are extraordinary. Private markets, there's $4.5 trillion of dry powder in private markets across all categories right now. So that $200 billion sounds large. We think it's actually just the early stages.
I think the good thing about the private markets as it relates to this multi-year issuance cycle is that we do expect the private markets will provide some certainty of financing in the later years. The simple point is that credit markets work best in funding re-leveraging when it's quantifiable and there's an end in sight. That's not really the case with this AI build out. So we do see a large role for private markets here.

**Paul Sweeney** (2:29)
Who are the borrowers when a data center gets announced and gets built? Is the borrower the construction company?

**Amanda Lynam** (2:34)
So typically the borrower is an SPV that is separate from the hyperscaler. But I think what you are alluding to is something that we've noticed in our investor conversations is that a lot of investors are increasingly counting their data center exposure in their hyperscaler bucket. And so I think it further increases our view. This is actually something we outlined in April, that issuer concentration and market saturation constraints will be binding here.

**Tom Keene** (2:59)
Okay, to Paul's brilliant question, and your even better answer, is this visible accounting? Can fancy people like you or Frank Fabozzi actually go in and understand the balance sheets of this new debt?

**Amanda Lynam** (3:16)
You can if you're willing to look at 10Ks and 10Qs, which we do, and actually just using the hyperscaler universe, there's about 1.2 trillion of lease commitments for data centers. Of that, 700 billion is for data centers that haven't started yet. They haven't begun construction. So to your question, Tom, that's not yet reflected in traditional leverage metrics.
That commitment for a data center that hasn't begun isn't yet counted in the financials. Some rating agencies and many investors are adjusting that after the fact, but that it is possible to do it if you're willing to get into the financials.

**Paul Sweeney** (3:51)
So there are special borrowers here. But again, is it if I'm going to my credit officer, I'm getting approval for this loan. Can I tell them at the end of the day, Microsoft is backstopping this thing?

**Amanda Lynam** (4:02)
All of the deals are different and whether or not they're fully amortizing or how the guarantee works. Or for example, if there is a construction delay, who's on the hook?
I unfortunately can't paint it with a broad brush.
But I think what is most critical from our perspective is that there's a lot of focus on the hyperscaler debt issuance, but actually data centers like the ones you are referencing have represented more than 20% of AI related supply this year. So it's important to track the AI related issuance from the broader tech ecosystem, not just the hyperscalers. And it further increases that competition for capital.