Why is Nvidia stock falling 2% on Thursday?

Nvidia shares continued to fall Thursday as a sell-off in AI-linked stocks intensified. Concerns that higher interest rates could constrain AI investment, fresh data pointing to mounting US inflationary pressures, and updates from TSMC and Anthropic fueled reassessment of AI spending, financing, and potential returns on infrastructure investment.

Analysis

The key distinction is duration and exposure to the financing loop, not simply “AI stocks” as one trade. NVDA is more directly exposed to whether customers can sustain exceptionally large AI infrastructure budgets and show returns; TSM is exposed to the same demand cycle, but its broader foundry activity may provide a relative buffer. That does not make TSM immune: a durable cut to AI orders would eventually pressure both utilization expectations and the rationale for capacity investment.

Near term, rising yields can compress long-duration multiples before any earnings estimates change, so price weakness alone is not evidence that AI demand has rolled over. Over the next 1–3 months, the important validation is customer capex guidance and evidence that deployment is translating into monetizable workloads—not startup announcements or sentiment headlines in isolation. Over 6–18 months, weak returns on deployed infrastructure could feed back into financing availability and orders, amplifying downside across chip suppliers.

The contrarian risk is treating rate-driven de-rating as a demand collapse; if yields stabilize and major buyers maintain budgets, crowded bearish positioning could reverse quickly. The article supplies no yield move, valuation, order data, or specific TSM/Anthropic update details, so avoid sizing an outright directional bet from this signal alone.

Key Decisions for Investors

• Prefer a relative-value watch over an outright AI short: consider long TSM / short NVDA only if the pair's valuation and recent relative performance support the entry. The thesis is lower direct sensitivity to AI-spending sentiment at TSM, not immunity to an AI capex downturn; define risk at the pair level and reassess if TSM order or utilization commentary weakens.
• For existing NVDA exposure, avoid adding solely because the shares have fallen. Wait for stabilization in real yields and confirmation from large customers' capex guidance or NVDA demand commentary; a further rise in yields alongside capex reductions would argue for reducing exposure rather than averaging down.
• Track US inflation data and Treasury yields over days to weeks, then customer capex, order visibility, and evidence of AI workload monetization over the next 1–3 months. Falsification of the bearish financing thesis: yields retreat while customer spending plans and supplier demand indicators remain intact.
• Verify what the cited TSM and Anthropic updates actually disclosed before trading the headlines. If they contain no change in orders, funding, or deployment economics, treat the sell-off as positioning/rates-driven; if they reveal a sustained spending or financing pullback, reassess semiconductor exposure across NVDA and TSM.