Odd Lots: "AI Can Tell Us Something About Credit Market Weakness"

Podcast: Odd Lots (Bloomberg)
Hosts: Joe Weisenthal & Tracy Alloway
Guest: Dan Wurtman, Co-founder of Noetica AI
Date: December 4, 2025

Episode Overview

This episode dives into the intersection of credit markets and artificial intelligence, exploring how AI technology is reshaping the world of credit agreements, deal terms, and risk management. With recent concerns about structural weaknesses in credit markets—including private credit blow-ups and increasingly complex lending structures—the hosts are joined by Dan Wurtman, whose company Noetica AI benchmarks credit and M&A deal terms using language models. The conversation focuses on how AI is uncovering and quantifying otherwise opaque risks and trends in credit, the evolution of deal terms, and the growing "flight to fortification" among market participants.

Key Discussion Points & Insights

1. Current Concerns in Credit Markets

• Context in Credit:

  • Ongoing worries about private credit, idiosyncratic defaults, and frauds that keep emerging.
  • Rising "creditor-on-creditor violence" and increasingly intricate financing structures.
  • Discussions of the "cockroach analogy"—the idea that when one problem surfaces, many more may be lurking (02:31, Tracy Alloway).

• "Each [default or fraud] is special in their own way. But the worrying aspect is that they keep coming to light."
  —Tracy Alloway (02:31)

2. AI's Role in Credit Markets

• Potential for Large Language Models:

  • AI can analyze massive, complex deal documents more efficiently and consistently than human associates.
  • Tools like Noetica aim to benchmark agreements against the market, helping identify off-market terms and latent risks (03:19, Joe Weisenthal).

• Example Use Case:

  • AI could flag risk-laden clauses or unusual language that might otherwise go unnoticed, helping to prevent future regret or financial loss for advisors and investors.
  • "It would be really nice if you could upload a credit agreement to ChatGPT and just say, make sure there's nothing in here that would get me in trouble."
    —Joe Weisenthal (04:29)

3. What Noetica AI Does & the Evolution of Deal Terms

• Noetica's Mission and Origins:

  • Provides AI-powered software for real-time benchmarking of credit, M&A, and capital markets deal terms.
  • Born out of Dan Wurtman's experiences at BlackRock and Wachtell, where he noted the absence of any real database for deal term benchmarking (05:47, Dan Wurtman).

• Why Deal Terms Matter:

  • Deal terms are like "the plumbing" of transactions—think speed limits and traffic lights for capital markets deals.

  • A McDonald's analogy is used: sometimes you might get an extra chicken nugget by mistake, but in finance, accidental outcomes—like Citibank's accidental $900M transfer—can have massive legal and financial consequences and prompt industry-wide changes (09:05, Dan Wurtman).

  • "Deal terms are really the underpinning of the entire transactional system... The rules of the road." (08:12, Dan Wurtman)

4. Emerging & Classic Deal Terms: Trends and Their Significance

• Key "Fortification" Trends:

  • Notable increases in lenders demanding stronger structural protections (e.g., "anti-PetSmart" terms, J.Crew blockers, anti-Certa protections).

    • Anti-PetSmart terms: Prevent companies from hollowing out collateral.
    • J.Crew blockers: Prevent transfer of valuable IP out of the reach of lenders (11:37, Dan Wurtman).

  • Detailed Data:

    • Anti-PetSmart terms: up to 28% of deals in Q3 (from 4% in 2023).
    • J.Crew blockers: now in 45% of deals (up from 15% in 2023).
    • Anti-Certa protections: spiked to 84% of deals, the largest QoQ jump on record.
    • "We're calling it a flight to fortification... We can very precisely tell you the percentages of deals that are actually getting a lot of these structural protections..."
      —Dan Wurtman (11:20)

• Borrowers’ Side:

  • More economic flexibility in deals via EBITDA addbacks (e.g., cost savings addbacks now in 64% of deals).
  • Net short lender terms (excluding lenders betting against a company) reaching all-time highs (13%)(16:25, Dan Wurtman).

• Implication: Indicates widespread anxiety and risk allocation in the market—both sides are reinforcing their positions in anticipation of future distress.

5. Technology Behind the Analysis—Why "Just Doing a Control-F" Isn't Enough

• Complexity in Legal Documents:

  • Language evolves, clauses may have long-range dependencies, terms can be hidden in different sections.
  • AI now enables semantic analysis, not just keyword search, capturing nuanced shifts in legal language and allowing for precise tracking across time and documents (24:52, Dan Wurtman).

• Noetica's Knowledge Graph:

  • Database contains over a billion deal terms, mapped for precedent and context—enormous resource for comparative analysis in both transactional work and litigation (26:13, Dan Wurtman).

  • "AI for the first time can attribute, in particular, new language models can attribute more semantic meaning to phrases and language that was impossible with things like N-grams."
    —Dan Wurtman (24:52)

6. Structural Weaknesses & Blow-Ups: Real-world Examples

• Recent Credit Market Nervousness:

  • Spreads on risky (CCC-rated) debt trending upward.
  • Lien subordination terms—governing who gets paid first in a bankruptcy—are escalating rapidly, showing market participants bracing for distress (29:50–32:25, Dan Wurtman).
    • "We clocked that [lien subordination] term at 84% of deals in Q3, biggest quarterly jump we've ever seen..." (30:25, Dan Wurtman)

• First Brands Case:

  • Used extensive off-balance-sheet receivables financing, leading to confusion and eventual distress when much more debt was uncovered during disclosure (34:54, Dan Wurtman).
  • "One of the creditor's lawyers [said] that $2.3 billion just disappeared..." (35:58, Dan Wurtman)

7. Complexity & Circularity in AI-related Financing

• Opaque and Leveraged Deals:

  • Complex, circular financing increasingly common, especially for large tech and AI infrastructure.
  • Pizza Analogy:
    • Joe loves pizza (Meta = customer); Dan opens a pizza restaurant (SPV) funded 90% with borrowed money (Blue Owl = financier), but Joe (Meta) is only a contract customer, not an owner.
    • If something happens to the restaurant, Joe loses his only food source—mirrors the underlying risk of off-balance-sheet financing in AI/data center deals (38:09–42:04).

• Meta-Blue Owl Data Center Deal Example:

  • Similar to pizza analogy: Highly levered (90%), off-balance-sheet, with contingent guarantees that don’t show up in Meta’s financials.
  • Concern: Mismatched incentives and unmanaged risks due to asset immaturity and obscured leverage.
    (39:56–42:31, Dan Wurtman)

8. Will AI-Induced Deal Complexity Lead to More or Fewer Problems?

• AI as Both Solution and Complication:

  • AI helps manage and catch emerging risks but will also accelerate the creation of new, complex deal terms.

  • Lawyers and bankers will likely continue the “cat and mouse” game of structuring new loopholes or protections (47:52, Joe Weisenthal).

    • "AI will be used to come up with new deal terms and the cat and mouse game will continue forever."
      —Joe Weisenthal (47:52)
    • "We will end up with thousands and thousands of pages of term sheets that, like humans, are just physically incapable of reading. It has to be read by AI, probably literally."
      —Tracy Alloway (48:12)

Notable Quotes & Memorable Moments

• On Default Trends and Deal Terms:
  "Many funds did not get back that hundreds of millions of dollars and litigation ensued. But a deal term in credit deals called 'erroneous payment deal terms' started popping up in the market... As of last quarter, 90% of deals."
  —Dan Wurtman describing the Citibank/Revlon incident's impact (09:38–10:48)

• On the Psychological Shift:
  "We're calling it a flight to fortification... [protections like anti-PetSmart, J.Crew blockers] are the highest we've ever recorded."
  —Dan Wurtman (11:20–12:46)

• On Emerging Structural Risks:
  "What we see is creditors may be preparing their system for distress... Over the last quarter, we started seeing people and lenders obsessed with lien subordination terms, which is the term that governs who gets paid first when everything falls apart."
  —Dan Wurtman (29:50–32:25)

• On Structural Opaqueness:
  "What First Brands used is a lot of receivables financing facilities that weren't properly disclosed..."
  —Dan Wurtman (35:18)

• On AI's Underlying Challenge:
  "[AI] can attribute more semantic meaning to phrases and language that was impossible... so that when you look at a J. Crew blocker... it may be phrased a thousand different ways, but we can track that term over time."
  —Dan Wurtman (24:52)

• On AI and Future Market Cycles:
  "I fundamentally believe AI is a paradigm shift... But I also think that categories will have winners and losers. And when folks are racing to define a category... there may be more losers in the headlines than you're used to seeing..."
  —Dan Wurtman (43:32–44:42)

Timestamps for Key Segments

• Credit Market Risks, Defaults & Fraud: 02:31–04:18
• AI’s Potential in Credit Documentation: 03:19–04:41
• Introduction of Guest & Noetica AI: 05:18–08:00
• Citibank/Revlon Deal Term Fallout: 09:05–10:48
• Trends in Protective Deal Terms: 11:20–15:25
• How AI Goes Beyond Basic Search: 24:04–25:50
• Lien Subordination Surge & Market Psychology: 29:50–32:25
• First Brands, Private Credit Structural Issues: 34:54–36:58
• Pizza Analogy for Circular Financing: 38:09–42:04
• Meta/Blue Owl Off-Balance-Sheet Deal Explained: 39:56–42:31
• AI's Double-Edged Sword for Deal Complexity: 47:52–48:23

Final Takeaways

• The credit market is showing clear signs of nervousness: both lenders and borrowers are fortifying their positions via increasingly complex, protective deal terms—as demonstrated by hard data from AI-driven analysis.
• AI’s ability to process, benchmark, and draw precedent from unprecedented amounts of contract data could transform deal-making, risk assessment, and even litigation prep.
• However, the pace of innovation—both in AI and financial engineering—ensures that structural risks and the need for ever-more-sophisticated protective measures will persist, perhaps even accelerate.
• Ultimately, AI can't eliminate credit risk or human ingenuity in exploiting loopholes, but it can provide a more transparent map of the battleground as new risks emerge and evolve.

Recommended for anyone interested in:
Credit market structure, legal innovation, financial risk management, the impact of AI on traditional industries, or just enjoying the analogies of McNuggets and all-you-can-eat pizza.