Carlyle’s Thomas: AI Slowdown ‘More Likely Than Not’

Carlyle's Jason Thomas said the Federal Reserve faces enormous pressure to deliver a 25bp rate hike, reinforcing a restrictive-policy risk for markets. He also said an AI investment slowdown is more likely than not, signaling caution toward the AI-driven capital-spending boom and related asset valuations.

Analysis

The relevant transmission for CG is not a single policy move but whether higher-for-longer rates keep private-market exit multiples and leveraged-finance issuance constrained. A modest additional tightening is largely a discount-rate issue for growth assets, but it can materially delay realizations if it pushes syndicated loan and high-yield spreads wider; that would extend fund-duration marks and defer performance-fee recognition. Watch private-credit origination as the offset: wider spreads and bank retrenchment can support deployment economics, provided defaults remain contained.

An AI-capex deceleration would be most consequential through public-market positioning and private-growth valuation marks rather than CG's operating earnings in the next quarter. The crowded beneficiaries—semiconductors, data-center infrastructure and power equipment—are vulnerable to a lower terminal-growth assumption; a de-rating in those assets could reduce exit appetite across technology-adjacent buyout portfolios over the next 6-18 months. Conversely, lower AI infrastructure spending would ease equipment lead times and power bottlenecks, benefiting non-AI industrial projects and businesses facing elevated compute costs.

Consensus is likely to treat a 25 bp move as either irrelevant or uniformly risk-off. The more useful distinction is between companies dependent on refinancing and asset managers able to lend into wider spreads: private-credit managers can gain share in a controlled slowdown, but lose sharply if the same conditions produce a default cycle. The near-term trade signal is weak absent evidence of widening credit spreads, reduced hyperscaler capex guidance, or lower fundraising/realization expectations.

Key Decisions for Investors

• No directional CG trade solely on this commentary; maintain a watch item for CG if high-yield spreads widen above 450 bp or leveraged-loan defaults accelerate, as those conditions would challenge realization timing and private-credit credit costs over the following 1-3 quarters.
• If a policy decision produces a rate-driven equity selloff without a material widening in credit spreads, consider a 1-3 month relative long CG versus KKR or BX only after confirming that CG fundraising and deployment remain intact; the thesis is that deployment fees and credit opportunities offset modest mark pressure. Exit if CG cuts fee-related earnings guidance or reports a meaningful realization slowdown.
• For an AI-slowdown confirmation—hyperscaler capex cuts, weaker semiconductor bookings, or reduced data-center power demand—favor a 3-6 month defensive pair: short SMH versus long XLU. The trade captures the potential unwind of AI hardware duration while retaining exposure to regulated power demand; invalidate if major cloud providers reaffirm accelerating capex plans.
• Avoid adding broad AI-beta longs ahead of the next earnings cycle unless revisions to revenue backlog and capex guidance remain positive. A 10-15% correction in AI infrastructure leaders without estimate cuts would be a better entry point than reacting to generalized caution.