Investment & Market Intelligence

The Investor

The Signal

JPMorgan's software markdowns put four banks' private-credit loans under Fed review.

The mechanism is back-leverage, the polite name for banks lending to funds against the loans those funds already hold, so a bank markdown leaves the fund less to borrow when it refinances the same companies. Distressed leveraged loans are up nearly 90% in a year. Fund marks are the lagging number, and they will catch up as sponsor-backed software borrowers refinance through 2028.

In Play

  1. Private Credit Meets AI Disruption

    Semafor reports the New York Fed has been reviewing JPMorgan, Wells Fargo, Barclays and Morgan Stanley's loans to private-credit firms since spring. The review was prompted partly by JPMorgan marking down loans to software companies facing AI disruption. The question for your book is whether the software loans inside your private credit funds are marked as conservatively as JPMorgan's. Distressed leveraged loans are up nearly 90% in a year, per Bloomberg data cited by The Bear Cave.

  2. Compute Debt Rides on One Customer

    Lambda's backlog rose from $15B in June to $50B in September, and about $35B of it comes from Anthropic, TheSequence reports. Lambda added $1B of debt days before raising up to $4B at a $14.5B pre-money valuation. Any GPU-cloud exposure in your book now carries single-customer credit risk. Separately, SpaceX, Broadcom and Digital Realty announced $120B of bonds in one day, per figures cited by The Bear Cave.

  3. Amazon's Valuation Leans on Neocloud Multiples

    The Information's Martin Peers argues Amazon's businesses are worth more than $3T against a $2.8T market cap. The load-bearing piece is AWS, valued at $1T or more by reference to where CoreWeave and Nebius trade. That ties a megacap's valuation to the same neocloud comps your AI-infrastructure exits rely on. Q3 earnings season starts in a week or two, and AWS growth against $220B of 2026 capex is the test.

  4. Private AI Rounds Price Before Revenue

    TypeSafe AI raised about $870M at a $7.5B valuation in an a16z-led Series A, under four weeks after a ~$200M seed and with no revenue disclosed, TheSequence reports. Nous Research raised $90M at $1.5B, about 42x its ~$36M mid-September ARR. Treat neither as a comp for your pipeline until TypeSafe discloses revenue and Nous reports year-end ARR against its $100M target.

  5. Exit Paths Split Along National Borders

    China forced Meta to unwind its $2B Manus acquisition, and Manus's parent then raised over $500M from domestic investors at a reported ~$4B, TheSequence reports. DeepSeek is pushing its round toward ~$15B at a ~$75B valuation, with Tencent, CATL and Geely among backers, ahead of a possible STAR Market listing. For US funds, China-origin AI assets have lost the Western strategic exit. SoftBank is separately seeking up to $100B from Gulf investors, the FT reports.

Deep Dives

  1. The Bank That Marked AI-Hit Software First Is Down the Least

    Lender write-downs usually reach fund statements late, so sponsor-backed software refinancing through 2028 is where losses still hidden in private marks will surface.

    How a bank markdown travels to your software holdings The channel is back-leverage : loans that banks make to private-credit funds, secured by the loans those funds hold. When a bank writes down the software loans inside that collateral, it…

    3 action items

    ●
  2. Lambda's $50B Backlog Is Mostly a Bet on Anthropic Paying

    A compute contract is worth what its buyer can pay, and this buyer is cutting prices while data-center borrowers line up record debt against demand still to come.

    Why the backlog behaves like a credit exposure Lambda's pre-money valuation comes to about 0.29x its backlog , a multiple that looks cheap only on the assumption that the backlog turns into cash. Most of it is owed by one…

    3 action items

    ●
  3. TypeSafe's $7.5B Prices a Workload That Small Models Are Cheapening

    Strategic and mega-fund checks set private marks before revenue or year-end ARR could justify them, and open weights arrive first to test them.

    The workload under the valuation is getting cheaper TypeSafe's product, Jev, returns typed decisions with confidence estimates for the small judgments inside business processes: classify a request, pick a tool, escalate an exception. Those high-volume calls are exactly the workload…

    3 action items

    ●

The edition continues

Take the signal into the room.

Sign up or log in to read all 3 deep dives in full, plus the final take.

Read the full edition

Continue with LinkedIn