Northland downgrades chip stocks on AI spending, supply concerns By Investing.com

Northland downgraded Astera Labs, Intel, and Semtech to Market Perform, warning that semiconductor stocks are priced for perfection and face elevated risk over the next two quarters. The firm flagged declining AI infrastructure spending, 4,300% higher AI model training costs since 2020, and supply chain stress from Iran-related conflict, rolling blackouts, and water/electricity constraints in Taiwan. It also cited potential Super El Niño disruption and limited access to critical production resources for TSMC, implying further pressure on the semiconductor supply chain.

Analysis

This is less a broad AI selloff and more a repricing of the margin structure behind the infrastructure layer. The first-order hit is to the high-beta “picks-and-shovels” names tied to new capacity buildouts, but the second-order effect is a likely rotation from growth-at-any-price AI capex into vendors with recurring software, utilization-based pricing, or exposure to inference optimization rather than raw deployment volume. The market is still underestimating how quickly hyperscaler capex can go from “strategic expansion” to “working-capital discipline” once debt-funded spending collides with cash-flow constraints.

The more interesting pressure point is packaging and test, where supply-chain fragility can turn a demand slowdown into a margin air pocket. If weather or power disruptions tighten capacity in Southeast Asia and Taiwan, weaker order books won’t translate cleanly into better pricing; instead, you can get delayed shipments, inventory digestion, and punitive mix shifts that hit names like INTC and SMTC harder than the headline AI beneficiaries. TSM is relatively better insulated, but the setup still argues for slower older-node output and a possible temporary support for leading-edge pricing if downstream customers hoard critical inventory.

The contrarian view is that the market may be over-discounting 2027 AI spending when the nearer-term catalyst is not a collapse in demand but a change in monetization. Usage caps, higher-tier subscriptions, and token-based pricing imply the industry is trying to force productivity ROI, which could actually improve the durability of inference demand even as free usage shrinks. That means the pure-capex beneficiaries may be vulnerable now, while the companies enabling cost compression and model efficiency could outperform over the next 6-12 months.

Key Decisions for Investors

• Short ALAB and SMTC into strength over the next 1-3 weeks; best risk/reward is a tactical short tied to capex multiple compression, with upside limited if the market treats this as a temporary digestion phase.
• Use INTC as a relative short versus TSM via a pair trade: short INTC / long TSM for a 1-3 month horizon, targeting supply-chain and execution divergence if Southeast Asia disruption worsens.
• For options, buy 3-6 month puts on a semicap basket or on SMTC specifically; the thesis is that valuation de-rating can outpace fundamental revisions before any supply-chain relief shows up.
• Avoid chasing TSM downside here; if weather/power risks intensify, TSM may become a relative safe haven on foundry scarcity, so any pullback is better expressed through weaker second-tier packaging names.
• If you want AI exposure, rotate toward profitability and efficiency enablers rather than capacity builders; prefer names with pricing power or inference optimization exposure over the next 6-12 months.