US manufacturing activity rises to four-year high in May, S&P Global survey shows

U.S. flash manufacturing PMI rose to 55.3 in May, the highest since May 2022, but the detail was inflationary: factory input prices jumped to 79.5, output prices to 63.3, and overall business input prices to 64.0, all multimonth highs. The report says firms are building safety stocks amid war-related shipping disruptions in the Strait of Hormuz, while delivery times worsened and private-sector employment fell to a 21-month low. The data point to stronger near-term activity but rising supply-chain pressure and a further acceleration in inflation.

Analysis

The market is being forced to reprice a classic late-cycle stagflation impulse: supply shock plus inventory hoarding boosts near-term industrial activity, but it is a poor-quality growth signal because it is being funded by precaution rather than final demand. That creates a setup where margins get squeezed twice — first by higher input costs and longer lead times, then by demand destruction when customers push back on price increases. The second-order winner is upstream pricing power, not cyclicals with weak pass-through.

For semis and adjacent hardware, the most important read-through is not demand strength but timing mismatch. If customers are rebuilding buffers now, that pulls forward orders into the next few weeks or months and can create a misleadingly strong bookings print, followed by an air pocket once stocking normalizes. That is especially relevant for high-expectation names where multiple expansion already assumes uninterrupted AI capex; any hint that enterprise buyers are funding inventories instead of new projects should compress duration-sensitive valuations first.

The inflation impulse is more dangerous than the growth impulse because it tightens the policy constraint even if activity slows. With delivery times worsening and prices-paid accelerating, the odds rise that rate-cut expectations get deferred, which mechanically pressures long-duration growth and keeps real yields sticky. In that regime, the market tends to punish the most expensive winners even if their fundamental growth is intact.

The contrarian view is that this is not a broad pro-growth regime but a narrow repricing of scarcity: logistics, defense-adjacent suppliers, and select commodity-linked industrials gain relative strength while demand-sensitive software and consumer discretionary names lose pricing elasticity. If the shipping disruption eases or the conflict de-escalates, the inventory bid reverses quickly, so this is a trade on supply fear with a short half-life rather than a durable earnings upgrade.

Key Decisions for Investors

• Short SMCI into strength over the next 2-6 weeks via call spreads or outright short exposure; the stock is vulnerable to a post-stocking air pocket if AI OEM/customer purchases were pulled forward. Risk/reward favors a fade because any positive order print is likely already crowded and valuation is highly duration-sensitive.
• Trim APP on rallies or hedge with near-dated puts for the next 1-2 months; ad-tech tends to be a late-cycle spend bucket and higher input inflation plus weaker consumer demand can hit advertiser ROI fast. Use a collar if you need to keep core exposure, since downside can accelerate if real yields stay elevated.
• Long SPGI only as a relative-quality hedge, not a directional bet, and pair it against a basket of industrials with thin margins. If inflation data stays hot for another 1-2 prints, data/benchmark franchises should hold up better than cyclical end-markets with no pricing power.
• Initiate a tactical long energy-transport / logistics hedge against broad industrial exposure for 4-8 weeks; delivery delays and inventory rebuilds support freight and storage demand even as broader growth softens. Keep stops tight because the thesis reverses quickly if shipping lanes normalize.
• If you want a cleaner macro expression, short QQQ versus long XLI for 1-3 months. The trade benefits from sticky rates and cost-push inflation, which usually hit long-duration tech harder than value/cyclical indices before the growth slowdown becomes obvious.