3 Reasons Why Nvidia's Monstrous Rally Can Continue

The article argues Nvidia's rally can extend through 2026-2027+, citing $650 billion in combined 2026 capex from the big four AI hyperscalers and Alphabet's planned $175 billion to $185 billion in 2026 capex with further increases expected in 2027. It also highlights Nvidia's valuation at about 24x forward earnings versus 21.8x for the S&P 500, framing the stock as reasonably priced despite recent gains. The piece is opinionated but supportive, with a constructive outlook for AI infrastructure spending and Nvidia demand.

Analysis

The market is still underestimating how sticky AI capex is once cloud operators have committed to multi-year model training and inference roadmaps. The second-order winner is not just NVDA, but the whole cluster of adjacent suppliers with exposure to power, networking, advanced packaging, and rack integration; those businesses can keep compounding even if GPU unit growth normalizes because the bottleneck shifts from chips to deployed capacity. In other words, the trade is increasingly about infrastructure intensity, not just semiconductor cyclicality.

The key risk is not that spending stops, but that mix changes. If hyperscalers push more of the budget toward in-house silicon, networking, or data-center power buildout, NVDA can still grow while gross-margin upside becomes more constrained and multiple expansion slows. The next real inflection point is 2027 guidance season: if commentary merely confirms continued growth rather than accelerating it, the current move can consolidate for months even without a fundamental miss.

The market seems to be pricing NVDA like a high-quality cyclical rather than a platform monopolist with multi-year visibility, which may still be conservative if capex keeps re-accelerating. But consensus is also likely too complacent about earnings durability: at this valuation, any evidence of digestion in orders or slower conversion of capex into revenue could trigger a sharp de-rating. The contrarian takeaway is that the better asymmetry may be in the picks-and-shovels around AI deployment rather than chasing the most crowded name after a strong run.

Key Decisions for Investors

• Stay long NVDA, but use a staged entry or add on 5-8% pullbacks; target 3-6 month upside if 2026 hyperscaler capex guidance remains firm, with a hard stop if customer commentary shifts from expansion to optimization.
• Prefer a basket long in AI infrastructure enablers versus outright NVDA concentration: long ANET/AVGO/SMCI (or equivalent infrastructure exposures) against short a semiconductor ETF sleeve if you want to isolate capex intensity without paying peak multiple risk.
• Express the 2027 continuation view with a call spread in NVDA dated 6-9 months out; this captures the market’s likely tendency to start discounting next year’s spend before official guidance, while limiting downside if the stock pauses.
• If you already own NVDA, hedge with a short-term collar into earnings or major customer capex updates; the near-term risk is not valuation in isolation, but any hint that orders are being pulled forward rather than sustained.
• Watch GOOGL as a confirmation trade rather than a primary long: if its capex commentary improves again, it supports the broader AI infrastructure tape; if it disappoints, reduce exposure to the whole chain because NVDA sentiment is tightly coupled to hyperscaler messaging.