Carlyle's Thomas Says It's Hard to Be Bearish Amid Capex Boom

Jason Thomas of Carlyle said US policy and the AI buildout are shaping markets, while nominal GDP growth is being driven too heavily by inflation. He also noted the Fed has become so accommodative in relative terms that holding rates steady now reads as dovish. The comments are macro-focused and directional, but no specific policy action or market-moving data was disclosed.

Analysis

The more important implication is that the macro regime is drifting toward a “higher-for-longer but easier-than-feared” equilibrium: policy may stay restrictive in name, yet real conditions can still loosen if inflation decelerates faster than growth. That matters for factor leadership because duration-sensitive assets no longer need outright cuts to re-rate; they just need the market to stop pricing additional tightening. In that setting, the first beneficiaries are the highest-beta beneficiaries of AI capex and financing access, while the losers are businesses that rely on cheap working capital and stable input costs.

The AI buildout creates a second-order squeeze on the ecosystem: hyperscalers and model leaders likely keep winning share, but the supporting stack can become a classic overcapacity trap if capital spending outruns monetization. Suppliers tied to data-center power, networking, and thermal management can still do well, but the risk shifts from demand to margin compression as procurement gets more competitive and customers demand price concessions. The market is still underappreciating how quickly “AI growth” can become a capex arms race that benefits the few with balance-sheet strength while punishing anyone funding expansion at the wrong point in the cycle.

The inflation point is more nuanced than it looks: if nominal GDP is being inflated rather than broadened by real activity, cyclicals with high operating leverage may disappoint even if top-line prints remain healthy. That is bearish for broad industrial beta and more supportive of quality growth, cash-rich software, and infrastructure-like beneficiaries that can pass through costs. The contrarian risk is that investors are too eager to fade any hawkish rhetoric; if labor remains tight or energy resumes pushing headline inflation, the market can quickly reprice from “dovish hold” back to “restrictive hold,” which would hit crowded long-duration positions within days.

Key Decisions for Investors

• Add to quality AI infrastructure exposure on weakness: long MSFT/NVDA on a 3-6 month horizon, sized for upside if capex remains robust but hedged with a basket short of low-quality software names that need easy financing. Risk/reward improves if real rates drift lower without formal Fed easing.
• Short small-cap/high-yield beta proxies for 1-3 months: pair long large-cap AI beneficiaries versus short IWM or a high-yield credit ETF to express the idea that restrictive policy still pressures funding-sensitive businesses even if the Fed is not hiking. Stops tight if credit spreads begin compressing materially.
• Buy call spreads on semicap and power-infrastructure beneficiaries over the next 2-4 months, not outright calls, to capture AI buildout upside while limiting valuation risk if the market rotates away from capex winners. Favor names with backlog visibility and pricing power.
• Avoid chasing broad cyclical exposure into any nominal GDP strength trade until real growth improves; if you want exposure, use a pair long quality industrials / short rate-sensitive unprofitable growth rather than outright long cyclicals. The trade should work if inflation stays sticky but activity does not accelerate.
• Watch for a reversal trigger in CPI/PCE and payrolls: if two consecutive prints reaccelerate, reduce duration-heavy longs by 25-50% and rotate into energy or inflation beneficiaries. The market is vulnerable to a fast hawkish repricing because positioning is likely leaning on the idea that no move equals dovishness.