Stifel: US non-residential construction spending up 0.1% in March By Investing.com

U.S. non-residential construction spending rose 0.1% year-over-year in March, with total non-building spending up 5.1% and building spending down 3.0%. Data center construction surged 33.7%, driven by hyperscaler capex and AI-related investment, while manufacturing spending fell 17.4% and office spending dropped 7.9%. Stifel said warehouse softness is easing and could improve in 2026, supporting demand for construction materials.

Analysis

The likely near-term winner is not just Intel, but the broader domestic semiconductor supply chain that gets pulled into a strategic onshoring narrative. A preliminary Apple partnership would validate Intel’s foundry credibility far more than its standalone product roadmap, which matters because customer proof is what narrows the valuation gap versus TSMC and Samsung over multiple quarters. For Apple, the strategic value is optionality: even a modest allocation to U.S.-based capacity reduces concentration risk and gives management leverage in future negotiations with Asian foundry partners.

The second-order effect is competitive pressure on contract manufacturers and tool vendors. If this is a multi-node, multi-year arrangement rather than a pilot, expect incremental demand for U.S. semiconductor capex, advanced packaging, and metrology equipment, with the biggest beneficiaries likely appearing first in order books rather than reported revenue. The main loser is not a direct peer stock so much as the market’s assumption that Apple’s supply chain remains stable and externally optimized; any diversification premium embedded into Apple’s margin profile could compress if domestic sourcing carries a persistent cost penalty.

The macro construction data reinforces the same hidden theme: AI infrastructure is still the cleanest capex vector, while legacy industrial end markets remain weak enough to keep pricing power contained elsewhere. Data-center intensity supports select materials and power-infrastructure names, but the deceleration in manufacturing and office spend argues against broad cyclicals and suggests a barbell environment where AI-linked winners outperform but traditional industrials lag for several quarters. The contrarian view is that investors may be overestimating how fast domestic chip manufacturing economics can scale; if this deal is mostly symbolic or limited to legacy nodes, the excitement around Intel could fade before fundamental earnings inflect, creating a classic headline-to-hard-data gap.

Key Decisions for Investors

• Long INTC vs short SOXX or SMH for 3-6 months: the asymmetry is that any confirmed Apple volume can re-rate Intel’s foundry multiple quickly, while the downside is capped by already depressed expectations; cut if no follow-through supplier announcements appear within 60-90 days.
• Buy AAPL on pullbacks only, but hedge with a small short in TSM or AVGO if the market starts pricing domestic-sourcing margin drag: Apple gains strategic resilience, yet the stock can underperform if investors focus on higher unit costs rather than optionality.
• Add a tactical long in AMAT or LRCX into any confirmation of U.S. foundry expansion: equipment names get paid first through capex budgets, and the trade should work over 1-2 quarters before consensus revenue models fully reset.
• Pair long XLU or data-center power beneficiaries against short broad industrial cyclicals like XLI over the next 6-12 months: AI-driven utility load growth is more durable than general manufacturing spend, and the latest construction mix argues for persistent dispersion.
• Avoid chasing broad construction-materials exposure until warehouse and manufacturing spending stabilize: the better entry is on a 2026 setup after another quarter or two of weak legacy-end-market data creates a lower base and clearer inflection.