2 AI Stocks to Buy and Hold for the Next Decade

The article favors Nvidia and Amazon as long-term AI investments, citing analysts’ estimate that AI infrastructure spending could total $31.6 trillion through 2050. Amazon’s Q2 AWS revenue rose 37% year over year to $42.2 billion, while AWS operating income increased nearly 63% to $16.6 billion, or 60% of Amazon’s total; management said cloud demand exceeds capacity for 2026, with visibility into 2028, and backlog reached $496 billion. Nvidia’s stated opportunities include a $200 billion CPU market; it raised its quarterly dividend 2,400% to $0.25 and expanded its share repurchase program, though the article notes competition and custom-chip risks.

Analysis

The more important signal is the potential shift in AI economics from scarce accelerators toward lower-cost, integrated compute. If Trainium can deliver competitive performance at lower total cost, AWS can win workloads while reducing its dependence on NVIDIA; that is a competitive risk to NVIDIA even if aggregate AI spending keeps expanding. It may also benefit custom-silicon enablers such as Broadcom, while AMD competes for workloads where customers want alternatives. The key uncertainty is workload-level performance and customer adoption—not management’s price-performance claims alone.

For Amazon, capacity constraints can convert demand into backlog, but they also defer revenue and leave execution exposed to power, data-center, and chip availability. Backlog is not equivalent to near-term recognized sales; watch conversion and AWS operating income alongside growth. For NVIDIA, platform breadth may defend share, but expansion into adjacent markets does not by itself offset hyperscalers’ incentive to internalize chip design.

Near term, this article is unlikely to be a durable catalyst absent new evidence; its valuation comparison lacks a common growth, quality, and cycle-adjusted basis. Over 1–3 months, AWS growth, capex, and capacity updates matter. Over 6–18 months, utilization and custom-chip adoption will test whether AI infrastructure returns justify spending. A capex pause or weak monetization could reverse the theme and pressure both names, with NVIDIA more exposed to accelerator demand and Amazon also exposed to broader cloud economics.

Key Decisions for Investors

• Avoid treating the article’s long-term spending estimate or NVIDIA valuation comparison as a standalone buy signal. Verify current valuation against forward earnings revisions and cash-flow expectations before adding exposure.
• Consider a modest relative-value position—long AMZN versus short NVDA—only for portfolios seeking to express a custom-silicon and cloud-integration thesis; size it as a hedge, not a high-conviction outright call. Reassess if AWS growth or operating income weakens, or if NVIDIA demonstrates durable share and pricing in custom-chip deployments.
• Track AWS backlog conversion, capacity delivery, and operating income, plus customer evidence of Trainium adoption and workload economics. Strong bookings without timely capacity conversion would weaken the Amazon thesis.
• Falsifiers: broad hyperscaler capex reductions, falling AWS growth or margins, persistent data-center/power constraints, or evidence that custom chips fail to win production workloads. These would challenge the bullish case for both stocks.