These are the top stock picks from UBS analysts heading into the third quarter

UBS named Vertiv and Nvidia among its Q3 2026 favorite stocks, highlighting AI “picks and shovels” tailwinds for Vertiv after its 92% YTD run and suggesting price-target upside of 18%+ from Wednesday’s close. For Nvidia, UBS cited the company’s outsized economic profit and forecast a >50% upside implied by the average analyst target, despite valuation assumptions that CFROI fades from 81% to 21% by 2030. UBS also flagged Ralph Lauren (+13% YTD) with ~7% additional upside expected from the average price target, reinforcing a broadly risk-on/AI-and-quality tilt in its basket.

Analysis

This looks less like a fresh information set and more like validation of an already-owned factor: secular AI capex plus a handful of self-help compounders. The alpha is in separating businesses that monetize physical bottlenecks from those that merely ride the narrative; VRT has more direct leverage to power, cooling, and data-center buildout constraints, so the real second-order beneficiaries are electrical/power names and data-center infrastructure landlords, not just semis. The risk is that the market is already paying for sustained scarcity, so any pause in hyperscaler spending would hit the equipment side faster than the chip side.

NVDA remains the higher-quality structural long, but it is also the most crowded expression of AI. The market is still discounting a post-2027 normalization, so the upside is not about proving demand exists; it is about whether the next platform transition preserves pricing power and prevents margin erosion. That makes NVDA sensitive to guide quality and gross margin commentary over the next 1-3 quarters, with the main falsifier being any sign that enterprise/custom ASIC substitution or supply normalization compresses economic profit faster than expected.

RL is the cleaner contrarian among the list because the thesis is self-help rather than macro beta: if pricing discipline holds, the multiple can stay supported even if discretionary demand cools. The consensus risk is underestimating how much of that improvement can persist without promotional pressure returning; if discounting comes back, the stock de-rates quickly despite decent earnings. KO, BSX, TGT, and SPOT read as lower-signal names here; the edge is weaker and the right move may simply be to wait for earnings or guidance rather than chase the UBS basket.

Key Decisions for Investors

• Long NVDA via a 3-6 month call spread on pullbacks, not strength; target a 1.5-2.0x premium payoff if Rubin-related estimate revisions keep margins intact. Falsify on any guide that implies gross margin compression or mid-teens revenue growth deceleration.
• Own VRT only on a 5-8% retracement or as a shorter-dated momentum trade; the stock is priced for continued hyperscaler capex acceleration, so upside is mostly in estimate revisions. Consider pairing long VRT against a weaker AI hardware name such as SMCI if you want to isolate quality infrastructure exposure.
• Long RL vs short XLY as a relative-value consumer-quality trade over the next 1-3 quarters. The thesis works if brand elevation and reduced discounting persist; cover if promotional activity returns or wholesale/China commentary weakens.
• No fresh trade in KO, BSX, TGT, or SPOT from this note alone; treat them as watch items until there is a catalyst in margin or guidance rather than an analyst upgrade-driven move.