CNBC Daily Open: Beware AI naysayers, the 'Hoax Buster' is here
Source: CNBC

AI-linked stocks sold off after President Trump rejected calls to slow frontier-AI development while Cohere CEO Aidan Gomez warned that AI models could become the most potent cyber weapon ever created; Micron, Intel, Marvell and Applied Materials each fell more than 4%, Nvidia declined about 3%, and SK Hynix's U.S. listing dropped 7%. Oil rose as Saudi Arabia's East-West pipeline remained closed, lifting WTI 1.34% to $101.39 per barrel and Brent 1.02% to $105.68. The 10-year Treasury yield briefly reached 5.014%, its highest since October 2023, as futures markets priced a 92% probability of a Fed rate hike following above-target inflation.
Analysis
The AI selloff is more likely a duration-and-positioning reset than a change in hyperscaler demand over the next quarter. A permissive federal posture reduces the probability of an immediate U.S. capex constraint, preserving the revenue runway for NVDA, MRVL and AMAT; however, a 5% Treasury yield raises the discount-rate hurdle for long-duration semiconductor multiples. Near term, crowded AI hardware exposure remains vulnerable through the Fed decision and any further yield backup, with MU and SKHY carrying the greatest downside beta because memory pricing expectations already embed an unusually strong HBM cycle.
The underappreciated consequence is a shift in AI spending mix: model capability concerns should pull incremental enterprise budgets toward identity, endpoint, cloud workload protection and software-supply-chain security rather than reduce total AI spend. PANW, CRWD, ZS and TENB are cleaner beneficiaries than frontier-model vendors because cyber risk monetizes through recurring security controls and compliance budgets; this can persist over 6-18 months even if semiconductor capex normalizes. Conversely, any material cyber incident linked to autonomous vulnerability discovery would create a regulatory risk premium for AI infrastructure names, potentially delaying enterprise deployments despite continued data-center construction.
MDLZ's manufacturing localization is strategically positive but too small to alter consolidated earnings near term. The real sensitivity is whether lower landed costs can protect Southeast Asian pricing and shelf availability while cocoa, energy and FX remain volatile; this is a 12-24 month margin-defense initiative, not a standalone catalyst. Higher oil and rates are a more immediate headwind to global consumer multiples and to AI valuation than the AI-policy debate itself.
Contrarian view: markets may be conflating calls for AI safeguards with a demand slowdown. Unless customers or hyperscalers explicitly reduce deployment budgets, regulatory attention is more likely to reallocate spend toward security and governance than impair compute demand. The thesis fails if the Fed signals multiple additional hikes, the 10-year sustains above 5.25%, or hyperscalers cut 2026 AI capex guidance.
AllMind Terminal
AI-powered research, real-time alerts, and portfolio analytics for institutional investors.
Request TrialMarket Sentiment
Overall Sentiment
mildly negative
Sentiment Score
-0.35
Ticker Sentiment
Key Decisions for Investors
- Use the next 1-3 months to run a relative-value AI-security trade: long PANW or CRWD versus short SMH, sized beta-neutral. Target a 10-15% relative return if security-budget reallocation emerges; exit if PANW/CRWD billings weaken or the 10-year yield falls below 4.5%, which would restore semiconductor duration leadership.
- Do not add directional exposure to MU or SKHY before the Fed and the next memory-pricing datapoints. Maintain or initiate a tactical short MU versus long NVDA for 4-8 weeks: MU has greater sensitivity to HBM pricing and inventory expectations, while NVDA retains the stronger software/ecosystem moat. Cover if DRAM contract-price estimates accelerate materially or MU raises gross-margin guidance.
- Buy NVDA only on a yield-driven drawdown rather than chase policy headlines: accumulate in tranches if the stock falls another 10-15% without a hyperscaler capex revision. Six- to twelve-month upside remains intact if AI capex holds, but use a 5.25% 10-year yield or a major customer deployment delay as risk limits.
- Keep MDLZ on a watch list rather than trade the facility announcement. Upgrade to a long only if upcoming results show Southeast Asia volume growth alongside gross-margin expansion; otherwise, cocoa inflation and currency pressure can absorb the logistical savings.
More News
- U.S. 10-yr yields surge past 5%: which stocks are most vulnerable?
- Trump says calls for more control on AI are a ‘SICK conspiracy’
- Asian stocks steady as AI rebound offsets oil, rate and bond-market pressure
- 'All eyes on the Fed': What NYSE insider Jay Woods is watching this week
- Not all tech is under fire in Monday's selling, while investors seek refuge in defensive stocks
- Trump, China Push Back on Calls to Slow AI