US Core Capital Goods Orders Surged in March by Most Since 2020

US core capital goods orders rose 3.3% in March, the biggest increase since mid-2020, after a revised 1.6% gain in February. The report points to a yearlong stretch of solid business equipment investment, supported in part by spending on artificial intelligence. The data are economically supportive but likely to have limited immediate market impact.

Analysis

The signal here is less about a single print and more about the durability of capex breadth: AI-linked investment is no longer confined to a narrow set of hyperscalers, it is spilling into networking, power, storage, cooling, and factory automation. That creates a second-order beneficiary set across semis, industrial automation, electrical gear, and datacenter infrastructure, while raising the bar for suppliers with long lead times and pricing power. The market should expect the strongest earnings revisions to show up in names that sell the picks-and-shovels of compute, not necessarily the GPU designers themselves.

The risk is that this remains a front-loaded spending cycle rather than a clean demand supercycle. If AI monetization lags the current buildout, customers may eventually pause orders, and the first places that typically see it are shorter-cycle industrials and component suppliers with less contractual visibility. A slowing in forward guidance over the next 1-2 quarters would matter more than any single macro print, because the equity market is already embedding a continuation of elevated capital intensity.

Consensus may be underestimating how inflationary this is for power, grid equipment, and specialized manufacturing capacity. The bottleneck is increasingly not chip demand but the physical ecosystem required to deploy it, which can sustain pricing power for select industrials even if tech multiples compress. The contrarian angle is that “good” capex can still be bad for margins in adjacent sectors: rising input costs, tighter labor markets in electrical installation, and longer equipment lead times can pressure non-AI industrial end markets and make broad cyclicals look falsely healthy.

Key Decisions for Investors

• Long semicap equipment and AI infrastructure basket over the next 1-3 months: favor AMAT/LRCX/ANET on any 3-5% pullback, with a view that order momentum and backlog conversion remain intact; risk/reward is attractive as revision upside should outpace multiple compression if the capex cycle extends.
• Pair trade: long VRT or ETN vs short a broad industrial ETF (XLI) for 2-4 months. The thesis is that datacenter power and thermal management capture the next leg of spend while general industrials are more exposed to margin pressure from rising input costs.
• Buy calls on SMH or SOXX into the next earnings season, 1-2 month tenor, to express upside convexity to continued AI-driven capex without taking single-name execution risk. Stop if hyperscaler commentary turns cautious on 2025 spending.
• Short-select short-duration industrial credit or hedge using IWM puts if macro data begin to roll over: the risk is that small-cap cyclicals are most exposed to a delayed capex pause and weaker financing conditions if AI spend normalizes.
• Avoid chasing broad-tech beta here; instead prefer long/short relative value within tech: long network/power beneficiaries, short hardware names with low switching costs and weak pricing power if the market starts pricing a capex air pocket 6-9 months out.