Ozan Tarman: AI Rally Still Has Room to Grow

AI companies are described as having considerable room for further expansion, supported by strong revenues and positive earnings. The article also notes that inflation and interest rates appear more predictable, which could improve the investment backdrop, though sentiment remains a key risk. Overall, the piece is constructive on AI fundamentals but is mainly analyst commentary rather than new market-moving information.

Analysis

The market is still pricing AI as a long-duration growth theme, but the more important inflection is that the burden of proof is shifting from narrative to cash conversion. That tends to favor the highest-quality platform winners with durable gross margins and real operating leverage, while low-quality “AI-washed” names usually lag once investors start discriminating between model access and monetization. If capital stays available and rates remain stable, the next leg is likely driven less by multiple expansion and more by earnings revisions, which is a healthier and more durable setup.

Second-order, the winners are not just the obvious GPU and cloud beneficiaries; the bigger opportunity may sit in the picks-and-shovels layer where demand is still underappreciated: power, networking, data center cooling, and storage. Those businesses can benefit even if AI application adoption slows, because the infrastructure buildout is front-loaded and sticky. The risk is that the market over-extends into the same narrow trade, creating crowding in mega-cap AI leaders and leaving the broader ecosystem under-owned until a rotation forces catch-up.

The main catalyst risk over the next 1-3 months is any re-acceleration in yields or inflation, which would compress duration-sensitive multiples first and fastest. Over 6-12 months, the bigger threat is a capex digestion phase: if hyperscalers continue spending but model monetization lags, investors may start demanding evidence of ROI, not just usage growth. That would hurt the second-tier software names most, because they rely on narrative momentum rather than direct balance-sheet proof.

The contrarian read is that the consensus may be underestimating how selective this tape becomes as fundamentals improve. Broad AI enthusiasm can still persist, but returns may come from relative value rather than beta: long cash-generative infrastructure and profitable platform leaders, short expensive application names with weak retention or unclear pricing power. In other words, the theme can be right even if the easy money in the most crowded names is mostly behind us.

Key Decisions for Investors

• Overweight XLK vs. IWM over the next 1-3 months: the AI trade should continue to outperform broad small-cap beta if rates stay contained; stop if 10Y yields break materially higher, as duration-sensitive growth leadership would likely unwind.
• Pair trade: long MSFT or NVDA, short a basket of unprofitable AI app names for 3-6 months; seek a 15-25% relative return if the market keeps rewarding earnings quality over story-driven multiple expansion.
• Add exposure to infrastructure beneficiaries like ANET, VRT, or CRWD on dips over the next 2-4 weeks; these names should capture second-order spending from AI buildout with better downside protection than crowded consumer-facing AI software.
• Sell upside calls on the most crowded mega-cap AI leaders into strength if implied vol remains elevated; the asymmetry is shifting toward earning power rather than multiple expansion, making overwriting attractive.
• Watch for a 5-10% pullback in AI software multiples as an entry point rather than chasing strength; the best risk/reward is likely to come after any rates-driven shakeout, not after a headline-driven breakout.