Nvidia Still Has 40% Upside – Here's the Math

The article argues the recent 3% S&P 500 pullback and 5% Nasdaq drawdown are normal volatility rather than a market meltdown, with Louis Navellier calling every dip a buying opportunity. It highlights Nvidia's 14th consecutive earnings beat, with EPS up 95% year over year to $1.87 and an estimated fair value near $300 per share, about 40% above current levels. The piece also points to an AI infrastructure shift from copper to optical systems, citing Marvell's 24% surge and Corning's nearly 13% jump after Jensen Huang's comments.

Analysis

The key signal here is not “AI is over” or “AI is back,” but that the market is shifting from a pure narrative trade to a two-layer trade: first-order earnings strength in semis and second-order bottlenecks in the physical stack. That favors companies with pricing power, constrained substitution risk, and identifiable capex linkages, while punishing names whose multiple expansion assumed uninterrupted hypergrowth. The near-term implication is that the best risk-adjusted longs may be in the picks-and-shovels layer, not the highest-beta compute owners.

NVDA still matters, but the asymmetry is narrowing: when a leader can outperform on fundamentals yet still be vulnerable to multiple compression, the cleaner expression is often via suppliers that benefit from the next leg of infrastructure spend. MRVL and GLW sit closer to the incremental dollars flowing into networking and optical transport, where the market can re-rate faster if hyperscaler capex rotates from GPUs to interconnect. That rotation is likely to be uneven over the next 1-2 quarters, creating multiple entry points rather than a single all-in move.

The contrarian read is that the crowd is still underestimating how much of AI economics can be preserved even if the chip trade cools. A 3%-5% market drawdown is not the issue; the real risk is a capex digestion phase that compresses the forward revenue visibility of the ecosystem. If hyperscalers signal slower near-term spending, high-multiple “AI adjacency” names can de-rate even while the secular thesis remains intact. That creates a good relative-value setup: own the physical infra winners and hedge the crowded momentum names.

APPF and NXST are a useful reminder that quality fundamentals can compound through volatility, but they are not the best way to express this theme today. The more actionable takeaway is to use any further market weakness to build exposure in infrastructure names with visible demand conversion, while keeping tight risk limits on the most consensus-heavy AI exposures.

Key Decisions for Investors

• Long MRVL / short NVDA as a 1-3 month relative-value pair: express AI capex rotation from compute to networking/optics; target 10-15% spread capture, stop if NVDA regains leadership on guide-up revisions.
• Add to GLW on pullbacks over the next 2-6 weeks: best expressed as a structural long for the optical transmission bottleneck; upside is driven by rerating plus durable order flow, with downside limited if AI capex merely pauses rather than reverses.
• Maintain a core NVDA long but hedge with short-dated call overwrites or call spreads: keep exposure to fundamental earnings power while reducing multiple-compression risk in a crowded name.
• Use weakness in APPF and NXST only as selective quality adds, not broad thematic exposure: these are lower-beta compounders, suitable for capital preservation if tech volatility deepens over summer.
• If hyperscaler capex commentary softens, rotate out of secondary AI beneficiaries first and into infrastructure leaders; that trade should be executed within days of earnings/guidance, not after the move.