5 Top-Ranked Momentum Stock Picks for Q3 2026 After a Fabulous Q2

U.S. equities closed Q2 2026 with sharp gains—S&P 500 +14.9% and Nasdaq +21.4%—and the article argues momentum should carry into Q3 via the AI trade, including a shift from chips into memory/storage and full data-center infrastructure. It highlights AI-driven capex at hyperscalers rising to $750B for 2026 (potentially $1T next year) and points to shortages lifting memory/storage prices, supporting picks including Micron (HBM demand; Zacks EPS growth forecast >100% and consensus earnings +16.3% in 7 days), Western Digital, and Microchip. It also ties large AI data-center buildouts to power/cooling demand and notes a May 28 U.S.-Iran MOU to extend the ceasefire for 60 days and discuss reopening the Strait of Hormuz as a potential geopolitical support.

Analysis

The cleaner read-through is not “AI is strong,” but that the AI bill of materials is broadening and becoming less discretionary. That favors the suppliers with real bottlenecks — especially MU, then CRDO — because scarcity in memory and high-speed interconnects tends to get priced through faster than it gets competed away. The second-order loser set is the margin pool on the customer side: hyperscalers, server OEMs, and integrated GPU vendors can absorb only so much component inflation before they either delay deployments or compress gross margin.

STRL is a different animal: it is a capacity-constrained execution story, not just a data-center beta. The key implication is that schedule compression can translate into revenue pull-forward and better capital turnover, which is why the upside is more visible over the next 2-4 quarters than over 2-3 years; the long-term risk is simply that the buildout pace normalizes and the market stops paying for scarcity. If this AI cycle moves from greenfield construction to utilization/optimization, STRL’s growth rate can decelerate faster than the current multiple assumes.

The contrarian view is that consensus is extrapolating a capex supercycle from a few quarters of tightness, but memory and storage are notoriously cyclical and can reverse abruptly once supply responds. For MU and WDC, the key falsifier is any slowdown in hyperscaler capex commentary or a flattening in contract pricing over the next 1-2 quarters; for CRDO, watch whether 800G/1.6T ramps are real demand or just pre-buying. WDC is the most tactical of the group: the market may be underestimating AI archival storage demand, but secular HDD skepticism caps the multiple, so the trade is better as a relative-value expression than a structural compounder.

Key Decisions for Investors

• Long MU into the next hyperscaler capex print; best risk/reward is on any 5-8% pullback, with thesis invalidation if DRAM/HBM pricing or FY27 consensus revisions roll over within 1 quarter.
• Pair trade: long MU / short SMH as a relative-scarcity trade; if AI capex keeps rising but component supply stays tight, MU should outperform broad semi beta over the next 1-3 months.
• Use CRDO on dips as a smaller-cap momentum expression, but prefer a call spread over outright stock if the tape has already re-rated the name; stop if optical/retimer revenue ramps slip one quarter.
• Long STRL vs short XLI or a basket of slower-execution industrials (e.g., MTZ/DY) for a 3-6 month relative-value trade; thesis breaks if data-center schedule compression normalizes or backlog conversion slows.
• Treat WDC as a tactical trade only: go long on weakness against a secular-storage skeptic basket, but take profits quickly if the market stops rewarding AI storage scarcity within 1-2 earnings cycles.