‘An investment thesis is not a religion’: $44 billion CIO warns you need to figure out how you might lose money, not just make it

Farallon Capital CIO Nicolas Giauque says AI is the 'single-most exciting area' in the market, but warns it could pressure SaaS and asset-light businesses tied to private credit portfolios. He cited BIS data showing SaaS-related direct loans rising from about $8 billion in 2015 to over $500 billion by end-2025, with Morgan Stanley estimating software at 26% of direct lending exposure and potential default rates rising to 8%, or as high as 13% in a worse AI-disruption case. The piece is mostly a strategic interview, but it highlights a growing AI-driven credit risk overhang and Farallon’s emphasis on restructuring opportunities into 2027 and beyond.

Analysis

The market is underpricing the lag between “AI threat” narratives and actual credit impairment. The selloff in software has already done some of the equity repricing, but the bigger second-order trade is in private credit: covenant stress will likely emerge first in refinancing windows, then show up in amendment fees, payment-in-kind toggles, and finally defaults. That creates a more attractive setup for managers with restructuring dry powder than for directional shorts in software, because the losses will be dispersed over quarters rather than realized in one clean air pocket.

The key competitive dynamic is that AI does not merely compress SaaS multiples; it changes the bargaining power between incumbent vendors, customers, and lenders. Vendors with sticky data/workflow advantages can repackage AI and defend retention, while asset-light vertical software with weak integration moats becomes vulnerable to price compression and higher churn. The hidden beneficiaries are not necessarily hyperscalers, but firms that monetize the disruption through tooling, workflow automation, and distressed-for-control situations once leverage becomes the constraint.

Consensus is likely overstating the near-term systemic risk and understating the medium-term dispersion. This looks less like a 2008-style shock and more like a rolling asset-quality problem concentrated in 2027+ vintage refinancings. That timing matters: it argues for patience on credit downside trades, because the best entry is after the market moves from narrative fear to realized EBITDA and covenant misses. In other words, the current phase is about screening for vulnerability; the monetization phase comes later, when lenders can no longer refinance the story.

For large-cap AI beneficiaries, the more important risk is not demand destruction but capex intensity and valuation asymmetry. If AI adoption broadens faster than expected, the best software names may still win; if not, the market will punish the entire cohort as if margins are structurally impaired. That creates a classic dispersion trade: own the platforms with distribution and proprietary data, short the subscale, usage-based SaaS models with high customer concentration and weak switching costs.

Key Decisions for Investors

• Long MSFT / short a basket of vulnerable SaaS names over 6-12 months: prefer platforms with data moats and enterprise distribution versus subscale per-seat vendors; target 15-20% relative outperformance if AI monetization proves uneven.
• Initiate a staged short in software credit risk via CDS or bond shorts on lower-quality SaaS issuers into refinancing windows over the next 12-24 months; risk/reward improves materially if default expectations reprice from low-single digits toward the 8%+ range.
• Overweight restructuring and special situations exposure now, with a 2027-2028 payoff horizon; the trade is long distressed paper / rescue financing / creditor-control situations that emerge after amendment fatigue becomes visible.
• Use put spreads on the most levered SaaS equity indices rather than outright shorts; this captures downside from multiple compression while limiting squeeze risk if AI productivity offsets arrive sooner than expected.
• Avoid chasing the current AI equity leadership at peak sentiment; wait for a 15-20% drawdown or a confirmed capex/monetization miss before adding, as the risk/reward is better after the market moves from optimism to proof.