1 Under-the-Radar AI Semiconductor Stock to Buy Hand Over Fist, According to Wall Street

Cohu's 2026 revenue is forecast to rise 23% to $558.5 million, accelerating from 13% growth in 2025, while its high-performance computing sales pipeline stands at $750 million. All seven Wall Street Journal-tracked analysts rate the stock a buy, with an average target of $57.43 implying 24% upside and a high target of $65 implying 40% upside. The article highlights strong AI-related demand, improving earnings trajectory, and continued stock momentum after a 160% gain over the past 12 months.

Analysis

The market is starting to re-rate the semiconductor equipment stack as a levered proxy on AI capex, but the cleaner second-order trade is not the chip designers — it’s the quality-control bottleneck. When memory and accelerator nodes get more complex, test yield and metrology intensity rise faster than wafer starts, so a mid-cap vendor with exposure to qualification cycles can see revenue inflect before the headline AI names do. That creates a lagged but potentially sharper operating leverage setup over the next 2-4 quarters if AI server buildouts remain on schedule.

The key dynamic is that this is a qualification story, not just an order story. Once a platform is embedded into a customer’s process flow, switching costs rise materially and the revenue stream can compound through node transitions, but the timing is lumpy and highly dependent on design wins moving from engagement to volume. That makes near-term upside more dependent on conversion rate than on macro demand, and it also means the stock can rerate quickly if a few large customers move from pipeline to backlog.

The consensus appears to be underestimating how much of the upside is already in the numbers for the next 12 months versus the optionality in 2027-2028. At current expectations, the market is pricing a transition from cyclical supplier to durable AI infrastructure beneficiary; if that transition stalls, multiple compression can be severe because earnings are still being normalized from a low base. The contrarian risk is that AI capex stays strong but qualification cycles elongate, pushing the revenue delta out by a year and leaving the stock vulnerable to a post-rally reset.

The broader takeaway is that this is a better expression of AI infrastructure breadth than a pure beta trade in mega-cap semis. If the memory upgrade cycle is real, equipment names tied to inspection and handling should outperform late-cycle chipmakers on the next leg because they monetize every increment of complexity, not just every unit shipped.

Key Decisions for Investors

• Long COHU on a 3-6 month horizon, using pullbacks as entry: risk/reward is attractive if the market is willing to price 2027 earnings visibility before the revenue shows up.
• Buy COHU Jan-2027 call spreads rather than common to express the 12-24 month qualification-to-revenue conversion thesis with capped downside and better convexity.
• Pair trade: long COHU / short a basket of large-cap semis with richer current multiples (e.g., NVDA, AMD) over the next 1-2 quarters to isolate equipment-leverage versus already-priced AI winners.
• If COHU rips another 15-20% before backlog converts, trim into strength; the stock’s biggest risk is not demand, but timeline slippage between pipeline and recognized revenue.
• Monitor order conversion and mix shift in inspection/metrology closely; if conversion stalls for one quarter, cut exposure quickly because the multiple can compress faster than consensus cuts.