Is Nvidia Stock Still a Buy After Returning to All-Time Highs?

Nvidia is still presented as a buy despite trading near all-time highs, with the stock at 24x forward earnings versus 21.6x for the S&P 500 and a Wall Street one-year target of $269 implying 35% upside. The article argues the AI build-out is still in early innings, citing Nvidia management’s estimate that global annual data center capex could reach $3 trillion to $4 trillion by 2030. The thesis is reinforced by the upcoming Rubin chip family and the view that growth can extend well beyond 2026-2027.

Analysis

The market is still underestimating the duration of the AI capex cycle, but the bigger second-order point is that NVDA is not just a semis trade anymore; it is increasingly a financing-and-infrastructure throughput trade. If hyperscaler build-outs keep stretching into 2030, the bottleneck shifts from model demand to power, land, and rack deployment, which means suppliers closest to the full-stack deployment layer can capture more of the economics than pure chip exposure alone. That favors ecosystem names with pricing power in networking, power management, optics, and deployment services, while creating lag risk for hardware-only competitors with weaker software attach or less control over the platform.

The consensus miss is that multiple expansion can continue even without heroic estimate revisions, because the market is likely still applying a 24-month runway to a multi-year earnings stream. But the main risk is not valuation compression; it is timing slippage. If data-center construction or utility interconnect delays push revenue recognition out by 2-4 quarters, the stock can de-rate abruptly even if the 2030 demand story remains intact. That makes the setup favorable for upside in the medium term but vulnerable to any evidence of order digestion or gross margin normalization.

The cleanest contrarian angle is that the most crowded long may be the simplest expression of the theme, while the under-owned winners are the picks-and-shovels around power density and inference efficiency. If Rubin meaningfully improves performance per watt, the beneficiaries extend beyond NVDA into suppliers that reduce total cost of ownership for hyperscalers; that can keep capex growth elevated while redistributing margin away from the chip OEM over time. In other words, the trade is not to short NVDA outright, but to rotate part of the exposure into the broader AI infrastructure basket before the market starts pricing in who actually captures the next dollar of spend.

Key Decisions for Investors

• Stay long NVDA on pullbacks over the next 1-3 months, but size it as a core momentum/quality hold rather than a tactical breakout trade; upside is still meaningful, but expect higher volatility around delivery or margin commentary.
• Add a basket long in AI infrastructure enablers (e.g., power, networking, optical components) over 6-12 months versus NVDA; the risk/reward improves if the market begins to question how much economics chip vendors retain versus the ecosystem.
• Use any post-earnings or conference-driven spike in NVDA to sell upside calls against existing longs (30-60 DTE), capturing premium while limiting the chance of giving back gains if the market reprices capex timing risk.
• Initiate a pair trade: long AI infrastructure suppliers / short a slower-growth semiconductor hardware peer with weaker software lock-in and lower exposure to hyperscaler spend; the thesis is relative share shift as deployment bottlenecks matter more than raw chip demand.
• Set a 2-4 quarter watchpoint for hyperscaler capex guidance and utility interconnect timelines; if those roll forward, trim NVDA by 20-30% and redeploy into the supply-chain names with earlier cash-flow conversion.