AI is scaring lenders from software

1. Key Themes

AI Is Creating a Structural Shift in Private Credit's Appetite for Software

Lenders are actively pulling back from software borrowers due to generative AI disruption risk. The retreat is broad: "Slightly more than half of those surveyed said they had reduced their exposure to software borrowers, including 21% who said they had significantly reduced their exposure and 6% who had stopped lending to them entirely." This is a durable repricing of sector risk, not a temporary cycle.

Rising Default Expectations Signal Private Credit Stress Ahead

The outlook for private credit defaults is sharply deteriorating. "69% of respondents said they expect default levels in private credit portfolios to increase in the next six months, up from 57% of respondents in Q1." The back-to-back quarters of worsening sentiment indicate a trend, not a blip.

Clean Energy Hits a PE Funding Record, Driven by Take-Privates

Institutional capital is flooding into energy infrastructure at scale. "PE funding for the clean energy market hit a quarterly record of $63.8 billion in Q1—driven by the take-private of power company AES Corp." This signals PE's conviction that energy infrastructure is a generational buying opportunity.

SpaceX's IPO Reframes How Public Markets Must Value Moonshot Companies

With SpaceX now publicly traded, investors need a new analytical framework. "The correct way to underwrite an Elon Musk-led company is not to take management timelines at face value, nor to dismiss them. The gap between the market's pricing of an announcement and the company's track record is where mispricings—and trading opportunities—form."

Agentic AI and Infrastructure Compute Are Drawing the Largest New VC Checks

Multiple large rounds this issue cluster around AI infrastructure and agentic applications: Sarvam raised $234M at a $1.5B valuation for Indian LLMs; Hydra Host raised $100M for a GPU compute marketplace; NewCore raised $66M for agentic AI security. The investment community is concentrating capital on the picks-and-shovels and application layers of AI simultaneously.

2. Contrarian Perspectives

SpaceX's Reliability Is Bifurcated—Hardware Is Credible, Timelines Are Not

The consensus view on Musk companies tends to be binary: either full believer or full skeptic. The data suggests a more nuanced split. "The underlying pattern shows the company's reliability is bifurcated: Hardware delivery is far more credible than timeline delivery. The company has achieved 100% of its tracked launch vehicle commitments and 100% of its human spaceflight commitments." But "Falcon Heavy arrived five years late; Crew Dragon's first crewed flight slipped three years." This framework—bet on the outcome, discount the date—is a more precise investing lens than most market participants apply.

SpaceX's Financial Projections Are Its Weakest Track Record, Not Its Ambitions

Investors often obsess over whether grand visions (e.g., Mars) are credible. The data shows the real risk is financial projections, which the market tends to treat as anchors. "In 2015, it targeted $12 billion in Starlink revenue by 2022 and $30 billion by 2025, yet it only hit $11.4 billion in 2025. In 2019, SpaceX pegged Starship development cost at $2 billion to $3 billion, but that cost has now run to $15 billion." Investors pricing in SpaceX's stated financial targets are likely to be consistently disappointed.

Some Private Credit Lenders Are Leaning Into Software Dislocation as Opportunity

While the majority are retreating from software lending, a contrarian cohort is moving the opposite direction. "Nine percent of respondents said they had increased their exposure, looking to capitalize on the dislocation." If defaults are priced in and spreads widen, this could prove to be an advantaged entry point for selective credit investors willing to do deeper underwriting.

3. Companies Identified

SpaceX

• Description: Private aerospace and satellite company, recently IPO'd
• Why mentioned: Subject of a PitchBook analysis benchmarking its milestone delivery track record as a framework for public market investors
• Quote: "With SpaceX finally publicly traded, its long list of promises is now front and center as stock market investors attempt to assess its value... 64% were eventually achieved, but only 17% arrived on schedule, with an average delay of 19.7 months."

Sarvam

• Description: Indian AI startup developing large language models
• Why mentioned: Raised $234M Series B at $1.5B valuation, led by HCLTech and Bessemer Venture Partners—a notable signal for non-US LLM investment
• Quote: "An Indian startup developing AI large language models, raised a $234 million Series B at a $1.5 billion valuation led by HCLTech and Bessemer Venture Partners."

Hydra Host

• Description: Miami-based GPU compute marketplace
• Why mentioned: Raised $100M Series A led by Kindred Ventures; represents the infrastructure layer of the AI boom
• Quote: "Developer of a marketplace for GPU compute, secured a $100 million Series A led by Kindred Ventures."

NewCore

• Description: Tel Aviv-based agentic AI security startup
• Why mentioned: Raised $66M led by Cyberstarts; illustrates the agentic AI security investment wave
• Quote: "An agentic AI security startup based in Tel Aviv, raised $66 million in new funding led by Cyberstarts."

Fin (formerly Intercom)

• Description: Agentic customer service AI startup
• Why mentioned: Acquired by Salesforce for $3.6B, one of the largest agentic AI exits to date; backed by Kleiner Perkins and GV
• Quote: "Salesforce is buying agentic customer service startup Fin (formerly known as Intercom), for $3.6 billion."

Tenstorrent

• Description: VC-backed AI chip designer
• Why mentioned: Qualcomm reportedly in talks to acquire for $8–10B, signaling consolidation in AI silicon
• Quote: "Qualcomm is in talks to buy VC-backed AI chip designer startup Tenstorrent in a deal valued between $8 billion and $10 billion."

Revolut

• Description: UK-based fintech/neobank
• Why mentioned: Case study in market disruption—"captured over 80% of Ireland's adult population as customers, illustrating how legacy banks weakened by post-2008 bailout restrictions left the market wide open for fintech disruption."

AES Corp.

• Description: US power company
• Why mentioned: Its take-private drove the record $63.8B PE clean energy quarter
• Quote: "PE funding for the clean energy market hit a quarterly record of $63.8 billion in Q1—driven by the take-private of power company AES Corp."

Centauri Health Solutions

• Description: Healthcare software company owned by Abry Partners
• Why mentioned: Subject of a $780M continuation fund, Abry extending ownership after 5+ years—signals conviction in healthcare software despite AI headwinds to the sector broadly
• Quote: "Abry Partners closed a $780 million continuation fund to extend its ownership of Centauri Health Solutions, a healthcare software company it acquired more than five years ago."

Ensemble Health Partners

• Description: Healthcare revenue cycle management company
• Why mentioned: Apollo-backed Thoreau Group in advanced talks to acquire for ~$12B—one of the largest healthcare services deals in the pipeline
• Quote: "Apollo Global Management-backed Thoreau Group is in advanced talks to acquire Ensemble Health Partners in a deal valued at about $12 billion."

Radical Numerics

• Description: San Francisco-based AI lab focused on biological research
• Why mentioned: Raised $50M seed led by Emergence Capital—notable for both its large seed size and bio-AI thesis
• Quote: "An AI lab focused on biological research, raised $50 million in seed funding led by Emergence Capital."

Roku

• Description: Streaming platform provider
• Why mentioned: Fox agreed to acquire for ~$22B in cash-and-stock deal—major media consolidation move
• Quote: "Fox agreed to acquire Roku in a cash-and-stock deal valuing the streaming platform provider at around $22 billion."

4. People Identified

Franco Granda

• Description: Senior Research Analyst at PitchBook
• Why mentioned: Author of the SpaceX milestone analysis piece
• Quote: "The correct way to underwrite an Elon Musk-led company is not to take management timelines at face value, nor to dismiss them."

Gwynne Shotwell

• Description: President of SpaceX
• Why mentioned: Rang the Nasdaq opening bell, marking SpaceX's public debut; central figure in the company's IPO milestone
• Quote: "SpaceX president Gwynne Shotwell, center, rings the Nasdaq opening bell Friday in New York City."

Zack Miller

• Description: Senior Private Credit Reporter at PitchBook
• Why mentioned: Author of the private credit survey piece on AI risks to software lenders
• Quote: "The private credit market is adjusting after a period of depressed dealmaking, as investors trim their exposure to software borrowers."

Hany Farid

• Description: Leading deepfake detection researcher
• Why mentioned: Cited as a signal of how far AI-generated media has outpaced detection capabilities—"the world's leading deepfake detective is now failing his own tests. Hany Farid watched the same missile video 100 times and still couldn't tell if it was real."

5. Operating Insights

For Software Companies: AI Risk Is Now a Lender Underwriting Criterion

Founders and CFOs should expect tighter credit availability. More than half of private credit lenders have already reduced software exposure. "21% who said they had significantly reduced their exposure and 6% who had stopped lending to them entirely." Software companies seeking debt financing need to proactively address AI disruption risk in their lender conversations—or face fewer options and higher spreads.

For Investors in SpaceX or Similar Companies: Track Milestones, Not Earnings

The article provides a direct operating framework for underwriting high-ambition, capital-intensive businesses. "Its stock price will be increasingly driven by performance against milestones and timelines instead of quarterly financials." Investors should build milestone scorecards—separating hardware delivery (high credibility) from financial projections and timelines (low credibility)—as their primary valuation tool.

For Operators Considering Continuation Funds: They Signal Conviction, Not Exit Failure

Abry Partners' $780M continuation fund for Centauri Health Solutions demonstrates that well-performing assets with long runways can attract dedicated capital beyond a fund's typical lifecycle—without forcing a premature sale into a difficult market.

6. Overlooked Insights

2021 Vintage VC Funds Are Deeply Underwater on Returns

Buried in the newsletter's benchmark data: the median IRR for 2021 vintage global VC funds is just 3.72%, with a median DPI of 0.00x. This means the majority of investors in that vintage have received zero cash back. Given that 2021 was a peak deployment year, this has significant downstream implications for LP re-up decisions, GP fundraising difficulty, and the secondary market for LP interests—yet it's presented with almost no commentary.

Deepfake Detection Has Effectively Broken Down

The brief mention of Hany Farid—described as "the world's leading deepfake detective" being unable to verify a missile video after 100 viewings—points to a verification crisis with massive implications for media, due diligence, legal evidence, and security. This is a potential investable theme (AI verification/provenance tech) receiving almost no attention in the current funding landscape covered here.