Table of Contents
Why Are Private Equity Firms Betting Against AI-Driven Code?
Apollo’s Strategic Shift: Shorting the Software Sector
Apollo Global Management, a heavyweight in the private equity world with over $900 billion in assets, has signaled a decisive shift in strategy. Throughout 2025, the firm actively reduced its exposure to enterprise software companies. The Financial Times reports that Apollo took “bearish” positions against corporate debt held by major software firms, including Internet Brands, SonicWall, and Perforce. These companies, owned by investment giants like KKR and Francisco Partners, previously represented stable growth.
This move marks a significant departure from the last decade, during which software investments dominated the $13 trillion private equity industry. Apollo’s leadership views the current software landscape as dangerously vulnerable to disruption from artificial intelligence. While AI presents opportunities, it introduces volatility that lenders detest. Marc Rowan, CEO of Apollo, summarized this risk at a recent conference: “Technological change will lead to massive disruptions in the credit market. I don’t know if this will benefit or harm enterprise software. As a lender, I’m not sure I want to find out.”
By closing these short positions after capitalizing on the volatility, Apollo demonstrated a protective approach. They fear AI will automate routine tasks in programming and customer service so rapidly that the underlying business models of these software borrowers could collapse.
The Investment Risk: Volatility and the Data Center Bubble
The hesitation extends beyond Apollo. Goldman Sachs has issued warnings regarding the sustainability of the current boom in AI data centers. The massive capital expenditure required to build this infrastructure may not yield proportionate returns if the practical utility of AI plateaus.
The core uncertainty lies in the labor market. Investors are struggling to predict whether AI will simply reduce the number of required programmers—increasing margins—or if it will commoditize software creation entirely, destroying the industry’s pricing power. For a firm like Apollo, which relies on predictable cash flows to service debt, this “binary outcome” risk is unacceptable.
The Developer Reality: The Efficiency Paradox
While investors worry about AI becoming too capable, a contrasting narrative is emerging from the technical front lines. The assumption that AI will write 90% of code by mid-2025—a prediction made by Anthropic CEO Dario Amodei—is facing practical resistance.
Hunter Leath, a developer at the San Francisco startup Archil, recently revealed that his team has moved away from AI code generation tools. This is significant because Archil operates within the AI data processing space; they are not Luddites, but early adopters. Leath, previously ranked among the top 50 users of the AI coding tool Devin, cited a critical efficiency bottleneck: the review process.
The team found that reviewing and debugging AI-generated code takes longer than writing the code manually. This “review tax” negates the speed advantage AI promises.
Conclusion: The Disconnect Between Hype and Utility
This developer feedback validates Apollo’s caution, though perhaps for a different reason. The market faces a dual threat. First, there is the financial risk that AI disrupts business models. Second, there is the operational reality that current AI coding tools may increase technical debt rather than productivity.
When top-tier developers reject AI tools because manual coding remains more efficient, the narrative that “AI will replace all software jobs” weakens. However, this inefficiency creates the very instability investors like Apollo seek to avoid. Whether the bubble bursts due to lack of utility or market disruption, the smart money is moving to the sidelines.