My Top 3 Predictions for AI Stocks in the Second Half

The article expects AI stocks to regain momentum in 2H as valuations reset and geopolitical conditions improve, with the S&P 500 up nearly 15% in Q2. For Nvidia, it calls for a growth “burst,” citing the Vera Rubin platform and potential CPU ramp into a ~$200B CPU market, including near-term guidance of almost $20B in standalone CPU sales for the year. For SpaceX, it forecasts more muted upside after its IPO surge (~+20% on day one), warning that heavy investment has already driven a $4.9B loss and could keep investor enthusiasm tempered.

Analysis

NVDA is the cleanest beneficiary, but the market mechanism is less about a single launch and more about wallet-share expansion into adjacent compute layers. If the company can credibly sell a CPU roadmap into agentic-AI stacks, it widens its addressable market and increases switching costs for hyperscalers; that supports multiple expansion more than near-term EPS alone. The key near-term risk is that investors front-run the story without seeing attach rates, so the stock can stall if the next print shows strong demand but no meaningful mix shift.

The bigger second-order effect is relative, not absolute: a broader AI rally reduces concentration risk and should lift semi/software breadth, but it also means leadership may rotate away from the most crowded megacaps. That helps less-owned AI enablers and adjacent tech baskets, while pressuring legacy CPU incumbents such as AMD and INTC if NVDA starts bundling compute more tightly. Over 6-18 months, that could compress the value of standalone CPU exposure unless they answer with ecosystem wins or pricing discipline.

SPCX looks like the more fragile setup. Private-market enthusiasm can survive one or two capital-heavy quarters, but a public-market listing forces focus on burn rate and funding needs, which usually caps upside unless there is a visible path to operating leverage. Contrarian view: consensus may be underestimating how quickly AI capital rotates from "story" names into proven cash generators; if that rotation persists, NVDA can still win while speculative satellite narratives underperform. Falsifier for the bullish NVDA view: any guide that implies CPU sales are incremental but immaterial, or a hyperscaler capex pause in the next 1-3 months.

Key Decisions for Investors

• Long NVDA vs short AMD for 1-3 months; thesis is that NVDA’s CPU optionality plus platform bundling will take share from standalone server-compute exposure. Stop if AMD shows better server attach or NVDA commentary suggests CPU is mostly a low-margin adjunct.
• Buy NVDA call spreads into the next product/commentary window; best risk/reward if the market is still skeptical and implied vol is not already pricing an adoption inflection. Exit on a post-event spike or if guidance fails to quantify CPU contribution.
• Short SPCX only on liquidity/borrow confirmation and only into strength; the trade is a capital-intensity and burn-rate reset as public investors force discipline. Cover if the company shows credible gross margin leverage or external funding comes at a premium.
• Overweight XLK over equal-weight tech for the next 1-3 months if breadth continues to improve, because broader participation should still leave the highest-quality AI cash generators in the lead. Falsify if AI leadership narrows back to only a few megacaps and breadth rolls over.