How to weather the current AI stock storm that we were all dreading

The commentary describes a pullback in data-center and semiconductor stocks amid concerns about AI funding, while Anthropic reportedly missed expectations despite a $100 billion annualized run rate. It warns that short-term rates, currently 3.75%–4%, could rise to 5.5%–6% if needed to bring inflation to the Fed’s 2% target, with oil near $100 per barrel adding pressure. The author says the portfolio holds 12% cash and is avoiding large net stock purchases, while noting that an end to the Iran war or fewer data-center restrictions could improve the outlook.

Analysis

AI infrastructure: funding is the marginal risk, not demand alone

The key transmission is from higher discount rates into project economics: expensive capital can defer data-center orders before end-user AI demand visibly weakens. That makes leveraged operators and long-lead-time suppliers more exposed than chip leaders with broader customer bases. NVIDIA’s endorsement may steady sentiment, but it does not remove refinancing, customer-concentration, or utilization risk at CoreWeave; treating chip collateral value as a blanket credit backstop is a category error. A prolonged build-out pause could also weaken pricing power for data-center components and shift spending toward power availability and efficiency rather than capacity.

Days: tape and rate sensitivity favor reducing gross exposure into rallies rather than buying the first drawdown. Over 1–3 months, watch real yields, oil, CoreWeave financing terms/backlog conversion, and hyperscaler capex guidance; state-level permitting outcomes are a slower, localized risk, not yet a national demand verdict. Over 6–18 months, sustained AI monetization would restore infrastructure demand, but likely with tighter underwriting and a premium for balance-sheet strength.

Contrarian: the market may be conflating a financing pause with an end to AI adoption. Conversely, a large annualized revenue claim from a private AI company is not proof of durable, profitable demand or data-center returns. The upside case needs evidence of customer economics, not just model usage. Thesis weakens if rates fall and capex guidance holds; it fails on the bear side if operators disclose cancellations, worsening utilization, or materially more costly refinancing.

Key Decisions for Investors

• Prefer a relative-value expression over a broad AI short: consider long NVDA against short CoreWeave on a relief rally, sizing conservatively because both retain AI-capex beta. Reassess if CoreWeave demonstrates improving financing access and utilization, or if NVIDIA’s own demand commentary deteriorates.
• Avoid adding to data-center component exposure solely on oversold conditions. For GLW and comparable suppliers, wait for evidence that customer orders and delivery schedules are holding; a sustained capex-guidance cut would invalidate the dip-buying case.
• Keep duration and gross exposure controlled while rates and oil remain upward risks. A sharp reversal in yields and oil, accompanied by stable hyperscaler capex plans, is the catalyst to rebuild quality AI exposure; renewed rate increases or weaker capex guidance argue for further de-risking.
• BNY may offer a less direct AI-capex exposure, but treat it as a conditional diversifier rather than a pure hedge: verify deposit trends, fee growth, and rate sensitivity before adding. No compelling standalone trade in CAH follows from this article.