AI Spend Has Been 'Critical' for US Economy, Says Torsten Slok

AI hardware was a key driver of August's increase in US import prices, while commodity volatility linked to the Iran war raised industrial-material costs at the border. The combination is increasing corporate input costs and AI-buildout expenses, potentially sustaining inflation pressure as the Fed assesses its rate decision. AI investment continues to provide support to US economic activity despite the cost headwinds.

Analysis

The investable issue is not a broad inflation resurgence but a potential bifurcation: AI capex can sustain nominal growth while simultaneously raising the delivered cost of servers, networking gear, power equipment and industrial inputs. That is margin-positive for bottleneck suppliers with pricing power—NVDA, AVGO, ANET, VRT, ETN and GEV—but unfavorable for enterprise adopters and data-center tenants whose AI monetization lags infrastructure spending. A higher-for-longer rate path would matter most for long-duration, cash-burning AI software and smaller cloud infrastructure names rather than the cash-rich semiconductor leaders.

Over the next 1-3 months, the key transmission channel is whether higher border costs appear in core goods inflation and corporate gross-margin commentary, rather than the headline import-price print itself. If input pressure persists while AI capex remains elevated, hyperscalers may protect spending by reallocating from labor, advertising, and non-AI infrastructure; this would reinforce the relative earnings outlook for AI hardware while weakening second-tier IT vendors. The contrarian risk is that hardware cost inflation reflects temporary freight, energy, or commodity distortions: a reversal in industrial-material prices would remove the Fed concern without impairing AI demand, favoring a renewed duration-led rally in software.

APO has no clean direct earnings sensitivity, but higher financing costs and commodity volatility can slow realization activity and reduce the valuation support for private-credit and private-equity marks. Conversely, resilient AI-related capital formation supports deployment opportunities in infrastructure credit and asset-backed financing. Treat any APO implication as second-order until management quantifies AI/infrastructure origination, portfolio exposure, or changes in realizations.

Key Decisions for Investors

• Maintain a 1-3 month quality-bottleneck basket: long NVDA/AVGO/ANET versus short IGV. The pair expresses AI capex resilience while hedging the risk that persistent inflation pushes real yields higher; reassess if 10-year real yields decline materially or hyperscalers cut capex guidance.
• Prefer VRT and ETN over data-center REIT exposure such as DLR for the next two quarters. Power and thermal-equipment suppliers can pass through constrained-component costs more readily, while REIT returns remain more rate-sensitive; invalidate on evidence of order cancellation, backlog deterioration, or falling pricing.
• Do not add directional APO exposure on this signal alone. Set an earnings watch for a rise in infrastructure-credit deployment and fee-related earnings alongside stable realization activity; absent that confirmation, commodity-driven volatility is more likely a valuation multiple headwind than an earnings catalyst.
• Use upcoming CPI/PPI and semiconductor-equipment lead times as triggers: a second consecutive acceleration in core goods costs supports the hardware-over-software pair; a cooling print combined with stable AI orders would favor covering the IGV short.