3 Best AI Stocks to Buy on This Pullback (SNDK, LITE, JBL)

The article argues that the pullback in AI infrastructure stocks reflects risk-off sentiment rather than weakening fundamentals, and highlights Sandisk, Lumentum and Jabil as potential buying opportunities. Reported results include Sandisk fiscal Q4 revenue up 372% year over year to nearly $9 billion, Lumentum revenue more than doubling to about $1.01 billion, and Jabil fiscal 2026 revenue up 21% to $36 billion; Jabil forecasts fiscal 2027 revenue of about $44.5 billion and AI-related revenue of $22.1 billion. The author favors Jabil, while noting risks from cyclical memory pricing, elevated valuations, and a possible slowdown in hyperscaler spending.

Analysis

The pullback is a duration-and-positioning test, not proof that AI infrastructure demand has weakened. The key transmission mechanism is whether hyperscaler capex converts into durable supplier pricing and returns on incremental capacity. If rates stay high, long-duration LITE is most exposed to multiple compression; a full order book into 2029 is not equivalent to guaranteed shipments, margins, or customer commitments. Optical architectures and sourcing can change before that backlog converts.

JBL is the cleaner risk-adjusted exposure: manufacturing complexity and customer outsourcing can support volume, but its lower-margin model means revenue growth alone is insufficient—watch core operating margins, free cash flow, and customer concentration. SNDK’s apparent cheapness is especially vulnerable to peak-cycle earnings optics: NAND supply additions or weaker pricing can overwhelm AI-driven bit growth. Western Digital is a separate company, not a proxy for Sandisk’s results; HDD substitution and storage mix are relevant sector checks, not direct read-throughs.

Over 1–3 months, guidance revisions, hyperscaler capex commentary, and LITE order/margin updates matter more than technical patterns. Over 6–18 months, capacity response and whether inference materially lifts storage demand determine cycle durability. The contrarian risk to the bullish thesis is that investors may be pricing scarcity as permanent while financing costs and power constraints delay deployments.

Key Decisions for Investors

• Prefer a staged JBL long over a broad AI-infrastructure chase; add only if guidance and core margins hold through the next update. The article’s cited $285 support/$320 resistance are conditional technical markers, not intrinsic-value levels. A break below support accompanied by weaker guidance invalidates the setup.
• Keep LITE on a confirmation watch rather than chase the premium: require the breakout retest to hold and verify that booked demand translates into shipments, capacity, and stable margins. A material order pushout or margin deterioration would falsify the bottleneck thesis.
• Treat SNDK as a tactical, high-beta cycle position, not a low-multiple value holding. Verify NAND contract pricing, industry supply plans, and data-center mix before adding; falling pricing or repeated estimate cuts would outweigh AI demand growth.
• Hedge thematic exposure rather than assume diversification among the three: all remain sensitive to hyperscaler capex and risk appetite. Reassess on capex reductions, rising financing costs, or broad downward estimate revisions; no trade is warranted solely because the stocks have pulled back.