Morgan Stanley Research put about $39 billion of software loans in its highest AI-disruption risk band, and about half of that already trades below 80 cents on the dollar, according to exhibits from the bank’s research reproduced by Exencial Research Partners in a post on X on 5 November. The exhibits cover 131 software borrowers with $247 billion of loans outstanding.
That is the other side of AI credit. Most of our coverage has followed the money borrowed to build AI, as in AI Runs on Credit Now. Who Holds the Risk?; these loans belong to software companies that AI could undercut.
Business Pill · Financial Conditions
An evergreen Business Engineer explainer on how financial conditions shape what companies can raise. It teaches the general idea only and does not discuss this research.
The key insight: Morgan Stanley’s exhibits put about $39 billion of software loans in their highest AI-disruption risk band; $19.8 billion of that, 51% by our arithmetic, is bid below 80 cents on the dollar, against none of the lowest-risk loans.
The universe. 131 issuers and $247 billion of loans outstanding, in six software archetypes (Exhibit 10).
The bands. Morgan Stanley sorts the loans into Green, Amber and Red bands; Exhibit 10 shows an average “moat risk”, “journey risk” and overall risk score for each archetype. Across the sector, 16% of outstanding loans fall in the Red band (Exhibit 10).
Prices. Exhibit 14 is titled “~$20bn of Red risk loans trade at distressed levels, vs. none in the Green bucket”. Of the Red loans, $19.8 billion is bid below 80, against $80.5 billion of the Green loans bid at 95 or above. The weighted average bid across the sector is 89 (Exhibit 10).
Where the risk sits. Front Office Applications have the highest share of loans in the Red band, at 32%; Cybersecurity has the highest share of outstanding loans trading at distressed levels, at 25% (Exhibit 10).
When it comes due. 38% of the sector’s outstanding loans mature in 2029 or earlier, and Exhibit 12 is titled “The 2028/29 maturities have high Red concentration (31%)”.
Dispersion. Exhibit 2 shows “greater price dispersion in the higher risk (higher overall score) Software loans”.
What Was Posted on X
๐จ MORGAN STANLEY JUST FLAGGED A $250 BILLION AI RISK HIDING IN THE SOFTWARE CREDIT MARKET.
While Nomura and Bernstein highlight the enormous opportunity in AI infrastructure, Morgan Stanley reveals the other side of the boom.
๐ด 16% HIGH AI disruption risk (~$40B) ๐ก 49% Moderate risk ๐ข 35% Low risk
But here’s where it gets interesting:
โข ~$20B of high-risk software loans already trade at distressed levels.
โข Average loan prices: 97 for low-risk vs. 73 for high-risk borrowers.
โข ~75% of high-risk exposure matures during 2027โ2029.
โข Weaker borrowers could face 550โ1,000 bps in additional refinancing costs.
โข Front-office applications have the highest share of high-risk loans at 32%.
These are AI vulnerability assessments, not default forecasts.
NOW CONNECT THE THREE REPORTS:
NOMURA: AI data centers are already generating substantial operating profits. Negative FCF largely reflects heavy upfront investment.
BERNSTEIN: OpenAI’s infrastructure commitments may be more manageable than feared, but compute oversupply could pressure neocloud economics.
MORGAN STANLEY: AI disruption risk is already influencing software debt valuationsโand refinancing pressure could intensify before earnings fully reflect disruption.
THE BIGGER PICTURE:
AI is creating extraordinary opportunities in infrastructure while threatening the economics of weaker software businesses.
The next big AI trade may not just be about who captures the billions flowing INTO AI infrastructure.
It may be about who loses pricing power, who struggles to refinance, and who gets left holding the debt.
AI isn’t just reshaping technology.
It’s starting to reshape credit markets.
— Exencial Research Partners (@exencial_rp) Nov 05 07:04
Exencial Research Partners posted this summary of Morgan Stanley’s software-loan research on X on 5 November, with an image of four of the bank’s exhibits. Our figures come only from those exhibits; the post’s other figures, such as average prices of 97 and 73, are its own and unchecked. The post calls the bands AI vulnerability assessments, not default forecasts.
Software loans in Morgan Stanley’s three AI-risk bands, split by bid price as a share of par, in $ billions outstanding. Data: Morgan Stanley Research (Exhibit 14; PitchBook | LCD), as reproduced by Exencial Research Partners. Totals and shares are our arithmetic.
The Red band in Exhibit 14 adds up to $39.2 billion ($7.1B + $5.8B + $6.5B + $19.8B), in line with 16% of $247 billion. $19.8 billion of it, 51%, is bid below 80 (our arithmetic).
The Green band adds up to $86.0 billion, about 35% of the total, and the Amber band to $121.5 billion, about 49% (our arithmetic). Exencial’s post gives the same 35% and 49%.
Amber also holds $19.0 billion bid below 80, so loans bid below 80 add up to $38.8 billion across the two higher bands (our arithmetic).
What the Post Adds
Exencial’s summary, which links the note to research from Nomura and Bernstein on AI infrastructure, also gives figures that are not among the exhibit values we transcribed: average loan prices of 97 for low-risk and 73 for high-risk borrowers, about 75% of high-risk exposure maturing in 2027 to 2029, and 550 to 1,000 basis points of extra refinancing cost for weaker borrowers. We could not check those against the note and do not use them in our numbers. The post describes the bands as “AI vulnerability assessments, not default forecasts”.
The framework says: “The traditional SaaS playbookโgood product, fair price, reasonable moatโis no longer viable.”
As we read it, Morgan Stanley’s “moat risk” score is a lender’s version of the same question. The framework does not discuss this research.
What Is Not Established
We have not read Morgan Stanley’s note itself, only the four exhibits reproduced in Exencial’s post; we found no press coverage of it. The exhibits do not name individual borrowers, and they do not say when the prices were taken. A low bid price reflects what buyers will pay today; it is not a default, and the bands are a risk assessment, not a forecast.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Morgan Stanley’s exhibits put 16% of $247 billion of software loans in their highest AI-risk band, and about half of that band is already bid below 80.
A note on sourcing. Every figure is from four Morgan Stanley Research exhibits (Exhibits 2, 10, 12 and 14) as reproduced in an image posted on X by Exencial Research Partners on 5 November 2026; we have not read the note itself and found no press coverage of it. Figures that appear only in Exencial’s text are attributed to the post and not used in our numbers. Totals and shares are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
