3 Stocks to Take Advantage of $1 Trillion in 2027 Capital Expenditures

The article argues that AI data center capex could reach $1 trillion next year, with Nvidia, Broadcom, and Sandisk positioned to benefit. Nvidia is cited as the industry leader with expected revenue growth of 81% in fiscal 2027, while Broadcom’s AI semiconductor business is projected to exceed $100 billion in revenue next year and grow 66% in fiscal 2026. Sandisk is described as a major beneficiary of SSD shortages, with analysts expecting 336% revenue growth in Q4 fiscal 2026 and 122% growth in fiscal 2027.

Analysis

The most important read-through is not that NVDA, AVGO, and SNDK all benefit from AI capex, but that the value chain is becoming more differentiated: accelerators capture the first-order spend, while custom silicon and storage capture the persistence of that spend. That matters because hyperscalers are increasingly optimizing for total cost of ownership, not headline FLOPS, which should widen the moat for vendors that can embed deeper into design cycles and lifecycle supply agreements. The market is still likely underestimating how much of 2027 earnings power will come from non-obvious bottlenecks such as advanced packaging, HBM adjacency, and SSD availability rather than just GPU unit growth.

The second-order winner is likely upstream suppliers with constrained capacity and long lead times, because every incremental dollar of capex forces buyers to lock in inventory earlier. That creates a setup where pricing power can persist longer than investors expect, but it also raises the odds of a sharp digestion phase if hyperscalers hit internal efficiency targets or throttle deployment after a year of heavy purchases. The timing mismatch is important: fundamentals may stay strong for 6-12 months even if sentiment turns before the actual 2027 revenue inflection.

The main contrarian risk is that consensus is extrapolating a straight-line capex ramp into straight-line earnings growth. If cloud buyers shift some workloads to lower-cost inference architectures, sovereign AI, or on-prem enterprise deployments, the mix could become less favorable for the highest-multiple names while still supporting the broader AI supply chain. That argues for preferring companies with tangible supply scarcity and contractual visibility over names whose upside depends on perpetual re-acceleration in spend.

Key Decisions for Investors

• Stay long NVDA on 3-6 month horizon, but prefer buying pullbacks or using call spreads rather than outright stock; the risk/reward remains favorable, but the trade is now more sensitive to any capex pause in early 2027 guidance updates.
• Overweight AVGO vs. NVDA as a relative-value expression for the next 6-12 months: AVGO has more visible custom-chip design wins and a broader software-like revenue base, which should cushion any GPU digestion phase.
• Use SNDK as the highest-beta expression of the storage shortage theme, but only tactically: long with a 3-6 month horizon, and pair with a partial hedge in broader semis because the stock is vulnerable to any normalization in SSD pricing.
• Pair trade: long AVGO / short INTC over 6-12 months. The gap should widen if custom AI silicon continues to take share from legacy CPU-centric datacenter spend and Intel remains stuck in a slower adoption curve.
• For risk-managed exposure, consider NVDA Jan-2027 calls financed by selling nearer-dated upside in semi basket ETFs; this preserves upside to the 2027 capex cycle while reducing premium bleed if the trade stalls before the next guidance step-up.