These are the most oversold stocks after a rough week on Wall Street

This looks less like an outright AI demand break and more like a funding-reset for the most capital-intensive parts of the stack. Companies with negative free cash flow and explicit financing needs are now trading on dilution risk and refinancing math, not on TAM narratives; that makes ORCL and SMCI the most fragile names if the market keeps demanding proof that AI capex converts to revenue within 1-2 quarters.

The second-order winner is not the headline AI platform names but the picks-and-shovels and balance-sheet-strong vendors that can survive a capex pause. TSM’s higher spending plan is a mixed signal near term: it pressures sentiment, but it also implies continued demand for the tooling ecosystem, so semicap proxies should eventually outperform the levered buildout stories. IBM’s disappointment matters because it weakens the “AI as incremental software uplift” trade and raises the bar for legacy IT multiples.

The contrarian setup is that the selloff may be over-discounting duration risk while under-discounting survivorship. If AI spend is merely rotating from speculative server buyers to hyperscalers with real cash flow, the market will likely punish financing-heavy names and reward the better-capitalized incumbents over the next 1-3 months. PYPL is the cleanest example of rumor-driven upside that can evaporate quickly absent formal action; CTAS is the opposite—proof that non-AI quality can still command a premium when investors get selective.