Nvidia Is Cooling, Microsoft Is Stumbling, and the Fed Won't Cut. Here's Why I'm Still Buying Artificial Intelligence (AI) Stocks.

The article argues that the recent pullback in AI stocks such as Nvidia and Microsoft is a buying opportunity for long-term investors, despite this year's weaker performance amid inflation, high rates, and geopolitical tensions. It highlights a shift from AI training to inference, which could benefit Broadcom, Intel, optical hardware suppliers, data center REITs, renewable power providers, and cloud AI infrastructure firms. The piece is primarily opinion-driven and unlikely to move markets on its own.

Analysis

The market is starting to separate AI monetization from AI capex. Training has been the headline trade, but the more durable cash flow pool is shifting to inference, which should re-rate the picks-and-shovels beneficiaries with recurring demand rather than bursty buildouts. That favors AVGO first, then the less obvious second-tier infrastructure names that monetize usage intensity rather than model wins: Intel’s CPU attach, opticals, colocation, and power infrastructure.

The underappreciated second-order effect is that broadening inference can compress the moat of the GPU-only trade. If marginal AI workloads increasingly run on custom silicon and general-purpose CPUs, Nvidia’s revenue growth can stay strong while multiple expansion stalls, especially if hyperscalers use procurement diversification as leverage. That also helps smaller cloud infrastructure providers like CRWV and NBIS if enterprises want capacity without committing to long-dated internal capex, but those names carry balance-sheet and utilization risk that can matter fast if demand slips.

Macro remains the main catalyst/risk bifurcation. A higher-for-longer rate backdrop hurts long-duration AI infrastructure multiples, but any easing in yields could quickly revive the trade because the sector’s terminal-value sensitivity is still extreme. The real near-term catalyst is guidance: if hyperscalers reaffirm capex despite macro noise, the market will likely reward AVGO, DLR, and optics names within 1-2 quarters; if they hint at optimization or delayed deployments, the entire AI complex could de-rate even if end demand stays intact.

Consensus is probably underestimating how much of the next leg in AI will be a supply-chain widening trade, not a single-stock winner take most. The better setup is to own the beneficiaries of AI diffusion and stay tactically cautious on the crowded high-beta leaders until earnings prove that inference monetization is offsetting slower multiple expansion.

Key Decisions for Investors

• Long AVGO into the next hyperscaler capex cycle; best risk/reward in the group because it benefits from inference mix shift and custom silicon adoption, with 6-12 month upside if demand stays broadening.
• Long DLR versus NVDA as a relative-value pair for 3-6 months; DLR should benefit if AI deployment keeps expanding into leased infrastructure, while NVDA faces multiple compression risk if capex broadens away from pure GPU demand.
• Add INTC as a tactical recovery trade over 1-2 quarters; if inference workloads keep migrating to CPUs and diversified AI nodes, the market may reprice it as an AI support beneficiary rather than a legacy drag.
• Use call spreads on LITE or GLW for a 2-4 month window; optics demand can surprise on capacity additions, but the position should be structured for upside without overpaying for cyclical volatility.
• Avoid chasing MSFT until sentiment stabilizes; if cloud AI spend normalizes without immediate monetization, downside from multiple compression may outweigh near-term revenue support.