US annual consumer inflation posts largest gain in three years as prices increase broadly

U.S. CPI rose 0.6% in April and 3.8% year over year, with energy prices up 3.8% and core CPI up 0.4%, reinforcing expectations that the Fed will stay on hold longer. The report pointed to higher gasoline, food, rents, and airfare, while Wall Street stocks fell, the dollar rose, and Treasury yields increased. The inflation spike is tied to the Iran conflict, supply-chain disruptions, and tariff pass-through, adding political risk for Trump and pressure on consumer spending.

Analysis

The market is pricing a regime shift from "inflation as transitory noise" to "inflation as an exogenous shock with policy spillovers." That matters for semis because the near-term issue is not demand for AI chips, but multiple compression: higher discount rates, slower capex approvals, and the risk that hyperscalers rephase AI infrastructure spending if energy and wage inflation squeeze free cash flow. NVDA is the cleanest expression of that vulnerability because its valuation embeds uninterrupted AI capex momentum; even a modest re-rating from 35x to low-30s forward earnings can erase a large amount of upside without any change in unit demand.

The second-order winner is not necessarily the obvious AI beneficiaries, but the equipment and memory ecosystem that can still grow through inventory restocking and mix shift. SMCI has more operating leverage to incremental server demand and can outperform if buyers pull forward orders ahead of further tariff pass-through or supply-chain stress, but it is also the most fragile to a capex pause because its revenue base is more cyclical and less recurring. APP is less directly exposed to the inflation print, but higher rates and weaker consumer sentiment can hit ad budgets with a lag; that makes it a better relative short only if the macro data start to bleed into traffic and conversion metrics over the next 1-2 quarters.

The key contrarian point is that the knee-jerk selloff in AI names may be overdone if investors are conflating headline inflation with actual AI demand destruction. Energy-driven inflation is usually a margin problem first, not a compute-demand problem, and large platforms may keep spending because AI infrastructure is now strategically tied to product differentiation. The bigger downside risk is that sticky inflation forces the Fed to stay restrictive longer, which compresses long-duration tech multiples even if fundamentals hold; that is a slower-moving but more durable headwind than any one CPI print.

Key Decisions for Investors

• Short NVDA into strength for a 4-8 week horizon via put spreads; the risk/reward is attractive if rates stay elevated and the market keeps de-rating AI duration, with upside convexity limited by still-strong secular demand.
• Pair trade long SMCI / short NVDA for the next 1-3 months to express relative AI capex sensitivity; SMCI should outperform on any near-term inventory pull-forward, but stop if hyperscaler guidance turns cautious.
• Avoid chasing APP here; if you want exposure, wait for 1-2 quarters of macro pass-through evidence before initiating. Higher rates plus weaker consumer sentiment create a delayed earnings-risk setup.
• For bullish AI exposure, prefer a staged entry only after real yields stabilize; buy NVDA on a 5-7% further drawdown if the selloff is rate-driven rather than demand-driven, with a 3-6 month horizon.
• Use call spreads on semicap equipment names instead of outright longs if you want AI exposure with less multiple risk; they capture continued capex while insulating against the Fed-driven de-rating in megacap tech.