Torsten Slok Shows Us How AI Is Eating the Entire US Economy | Odd Lots

Apollo chief economist Torsten Slok argues AI is now a major driver of the U.S. economy, citing data center buildout, the CHIPS Act-driven industrial rebound, and AI’s contribution to GDP. He also said nonfarm payrolls would rise, consistent with last week’s strong jobs data. His portfolio takeaway is that the old 60/40 stock-bond framework is giving way to an AI vs. non-AI split.

Analysis

The investable takeaway is not simply that AI is supporting growth, but that it is becoming the marginal source of demand across multiple capital pools at once: corporate capex, utility load, construction, and labor-intensive services around buildout. That creates a reflexive loop where stronger GDP printouts and payroll resilience validate higher earnings multiples for the AI complex while also keeping real yields firmer, which is a hidden headwind for duration-sensitive parts of the market. In other words, the trade is less about “AI winners” in isolation and more about the crowding-out effect on everything that depends on cheaper capital and softer macro data.

Second-order winners are the picks-and-shovels names tied to power, cooling, networking, and industrial automation, because the bottleneck is shifting from semis to physical infrastructure. The more interesting edge is that the data-center boom is likely to compress timelines for grid interconnects and backup generation, which can re-rate select industrials before revenue shows up in the broader economy. By contrast, sectors that rely on labor slack or falling rates to support margins—small-cap cyclicals, homebuilders, and levered growth—could lag if AI keeps inflating demand and keeping policy restrictive longer than consensus expects.

The contrarian risk is that the market is over-assuming a straight-line AI capex cycle. If hyperscaler spending normalizes or AI monetization disappoints, the “AI vs non-AI” portfolio framing can unwind fast because the macro support becomes more of a cost burden than a growth tailwind. The catalyst to watch over the next 1-3 months is whether payroll strength and sticky core inflation force the Fed to stay higher for longer, which would punish the non-AI long book even if AI leaders remain bid.

For positioning, the cleanest expression is a relative-value trade rather than a broad beta bet: stay long AI infrastructure beneficiaries while hedging via shorts in rate-sensitive, low-quality growth. The risk/reward is best where earnings revisions can compound for 12-18 months, but valuation is still anchored by near-term cash flow rather than distant TAM narratives.

Key Decisions for Investors

• Go long AMAT / ANET on a 3-6 month horizon; use pullbacks to add. The setup is favorable because incremental AI capex still flows through the supply chain before it shows up in end-demand data. Risk/reward: 2:1 if data-center spending holds, with downside capped by visible backlog.
• Pair trade: long QQQ / short IWM for 1-3 months. If AI is the marginal growth engine and rates stay sticky, mega-cap balance sheets should outperform levered small caps. Target 5-8% relative outperformance; cut if rates break meaningfully lower.
• Long VRT or ETN versus short a basket of rate-sensitive industrial laggards. This is a cleaner way to express the power-and-cooling bottleneck created by AI infrastructure buildout. Best entered on any post-print weakness, with a medium-term horizon of 6-12 months.
• Buy 3-6 month puts on unprofitable software names with long-duration cash flows. If the market re-prices the 'AI vs non-AI' split under higher-for-longer rates, these names should be the most vulnerable to multiple compression.
• For event risk, hold optionality around CPI/NFP releases via call spreads in XLK and put spreads in IWM. The macro linkage is tighter than usual: hot labor data helps AI capex via demand, but hurts the rest of the market through discount rates.