AI Infrastructure Spending Could Nearly Triple by 2029. Here Are 2 Stocks to Buy.

Statista projects AI infrastructure spending will reach $902 billion by 2029 (up from $334 billion in 2025), underpinning sustained demand for AI data-center hardware. Nvidia remains the dominant supplier — data-center sales were >90% of revenue and grew ~75% YoY, with the company reporting ~$120 billion net income on $215 billion revenue last year and citing ~$1 trillion in cumulative Blackwell/Rubin purchase orders through 2027. Iren is positioned on the power side with >4.5 GW secured, a $9.7 billion Microsoft contract, and a targeted $3.4 billion annualized run-rate revenue by end-2026 (about 10% of its secured capacity), while the company’s $13 billion market cap implies significant optionality if more hyperscaler deals are monetized. Key risks: data-center spending has historically fallen in waves (past drops >50% from peaks) and Iren faces financing/dilution execution risk despite founder-aligned ownership.

Analysis

The market is converging on a platform effect where vertically integrated suppliers that combine accelerators, high‑speed interconnects, and turnkey rack systems create sticky demand because they shorten deployment time and reduce integration risk for hyperscalers. That stickiness amplifies lead‑time power: customers will pay a premium for immediately available, validated stacks rather than cheaper point solutions, which pushes value upstream to firms that control either the stack or the critical inputs (power, switchgear, trained build teams).

A less‑obvious beneficiary set are the capital‑intensive enablers of sustained rack density — grid upgrades, fast‑dispatch generation, large step‑up transformers, and specialist EPCs — because permitting and procurement timelines for power equipment are measured in quarters to years, not days. Owners who can pre‑secure grid capacity convert that timing arbitrage into contractual leverage: shorter delivery windows reduce churn and increase switching costs for customers, allowing operators to capture higher rents per MW and exercise pricing discipline on hosting margins.

Key tail risks are demand cyclicality and a rapid improvement in software or model efficiency that materially reduces GPU compute per model — either can compress utilization and reverse pricing power quickly. Near term (0–12 months) watch order cadence and inventory disclosures; medium term (12–36 months) the critical catalysts are new hyperscaler deal announcements, large permit approvals for grid capacity, and any export‑control changes that re‑route procurement. Capital structure risk (equity raises to fund buildouts) is the principal company‑level hazard for balance‑sheet‑heavy operators and is the main compression vector for returns if financing costs stay elevated.

Key Decisions for Investors

• Long NVDA via a calendar call spread (buy 12‑month ATM calls, sell 3‑month OTM calls to fund) — directional play on continued platform lock‑in with defined downside if near‑term order cadence slows; target 30–40% upside over 12 months, max loss limited to net premium paid.
• Long IREN (IREN) equity with put protection (buy 12–24 month puts ~10–15% OTM) — capture optionality from secular demand for pre‑secured power while limiting dilution/financing tail risk; size as a thematic allocation (3–5% of thematic book), reward skew attractive if monetization accelerates.
• Pair trade: long DELL (DELL) hardware exposure / short INTC (INTC) legacy CPU exposure — trade the wedge between OEMs capturing GPU server volume and chip incumbents that may lose attach‑rate; use 6–12 month expiries (buy DELL calls + buy INTC puts) to keep directional exposure but cap carry.
• Event trigger rule: if a major hyperscaler signs a multi‑year campus power deal with any provider, buy the provider and related infrastructure names on a 6–18 month time horizon; conversely, if broad GPU order revisions show downticks in quarterly spend, reduce delta in NVDA position by 40% within 5 trading days.