Looking for AI stocks that are still cheap? Consider these two under-the-radar picks.

Saturna Capital’s Simon Kim and co-managers say the AI build-out has further to run and remain overweight AI despite market jitters about high valuations. They are seeking attractively valued stocks in related industries, including two tied to base metals; the excerpt does not name the stocks or provide valuation figures.

Analysis

The investable thesis is an indirect-capex rotation, not a clean AI proxy. Data-center and grid expansion can lift demand for copper and other base metals, but miners’ earnings still depend heavily on broad industrial demand, supply disruptions, costs and commodity prices. That creates a potential diversification benefit versus crowded AI leaders, but not a reliable hedge against an AI-capex slowdown. The second-order risk is that investors capitalize an AI narrative before incremental metal demand is material: efficiency gains, substitution, recycling and project delays could blunt the demand impulse, while new mine supply takes years to respond. The article excerpt does not identify the two companies or provide valuation, production mix or balance-sheet data, so relative cheapness and company-specific exposure cannot be tested. Near term, this is more likely a sentiment/rotation catalyst than an earnings catalyst. Over 1–3 months, validate the thesis against data-center and utility capex guidance, metal prices and inventory trends; over 6–18 months, watch whether project orders translate into realized volumes and pricing. The contrarian risk is that “AI-linked” becomes a valuation label on cyclical businesses whose dominant driver remains the global manufacturing cycle.

Key Decisions for Investors

• No single-name trade yet: obtain the omitted company names, then verify revenue and production exposure by metal, cost position, debt, valuation and sensitivity to realized prices before treating either as an AI beneficiary.
• Keep base-metal producers on a watchlist as a potential relative-value diversifier to high-multiple AI equities, not as a direct AI hedge. Consider a pair only if company-level valuation and commodity sensitivities support it; the excerpt supplies neither.
• For a 1–3 month confirmation signal, monitor copper and other relevant metal prices and inventories alongside hyperscaler, data-center and utility capex guidance. If infrastructure spending holds but metal indicators weaken, the demand-to-earnings link is not yet validating.
• Falsify the thesis if AI/grid project deferrals accumulate, relevant metal prices and inventories deteriorate, or the named companies fail to show improving volumes or guidance despite continued infrastructure investment.