PIMCO Warns of Cracks in Credit

PIMCO strategist Lotfi Karoui warned that cracks are emerging among the weakest credit borrowers and that an upcoming wave of AI-related debt could require substantially more selective underwriting. He emphasized downside protection and deal structure over attempting to identify the eventual AI winner, signaling elevated default and credit-selection risk in weaker issuers.

Analysis

The actionable signal is not broad AI de-risking; it is a likely widening in dispersion between issuers funding compute build-outs with internally generated cash flow and those reliant on incremental debt. Hyperscaler capex remains manageable while free-cash-flow conversion and net leverage hold, but the marginal AI infrastructure borrower faces a more punitive refinancing market if revenue realization trails depreciation, power costs, and interest expense. This favors credit selection over index beta during the next 1-3 months, particularly in lower-rated technology, data-center, and digital-infrastructure issuers.

Second-order pressure should surface in the private-credit and leveraged-loan complex before it is visible in headline investment-grade spreads. Aggressive EBITDA add-backs, delayed cash-interest structures, and loose portability provisions can turn a modest operating miss into a liquidity event; public BDCs with meaningful exposure to sponsor-backed software, telecom, and infrastructure borrowers are a liquid proxy for that risk. Conversely, power and grid suppliers with contracted demand may be insulated from an AI capex slowdown, since constrained interconnection capacity gives them better pricing power than commoditized data-center construction vendors.

The consensus risk is treating all AI financing as equivalent. A credit accident among speculative build-out vehicles could improve the competitive position and returns on capital of MSFT, GOOGL, AMZN, and META by reducing capacity added by weaker competitors. The thesis is falsified if high-yield spreads remain contained while new AI-linked issuance clears without materially higher coupons or covenant concessions, or if hyperscalers demonstrate AI revenue sufficient to stabilize capex-to-revenue ratios over the next two earnings cycles.

Key Decisions for Investors

• Prefer quality AI exposure through long MSFT / GOOGL versus a short basket of high-beta data-center and digital-infrastructure credit proxies; implement over 1-3 months only after identifying issuers with near-term maturities, rising interest burden, or weak cash-flow conversion. Target 10-15% relative downside protection in a credit-spread widening; exit if HY OAS tightens materially and refinancing calendars clear cleanly.
• Reduce broad high-yield beta via HYG or JNK hedges rather than selling investment-grade technology credit outright. Use a 3-6 month horizon: the likely catalyst is weaker-than-expected earnings guidance or a failed/repriced leveraged financing, while the hedge should be covered if HY spreads fail to widen after the next heavy issuance window.
• Underweight BDC exposure with concentrated software, telecom, and sponsor-finance books; use ARCC and OBDC as monitoring vehicles rather than blanket shorts. Escalate to a short only if non-accruals rise, NAV marks weaken, or management reports deteriorating interest coverage—data not supplied here.
• Do not short hyperscalers solely on AI-debt concerns. Their balance sheets can convert weaker competitors' financing stress into lower future capacity competition; reassess only if capex accelerates while AI monetization disappoints for two consecutive reporting periods.