Goldman Sachs Projects $700 Billion in Artificial Intelligence (AI) Capex This Year. Here's My Top Stock to Buy.

Goldman Sachs estimates $500 billion is likely — and up to $700 billion at peak levels — could be spent in 2026 to build AI infrastructure, though bottlenecks may constrain investment. Brookfield Renewable has ~13.5 GW of demand pipeline from Microsoft and Google, offers a corporate yield of 3.9% and partnership yield of 4.9%, has delivered ~5% annualized dividend growth historically, and targets 5–9% annual dividend growth going forward, positioning it as a long-term beneficiary of AI-driven power demand.

Analysis

The AI-driven spike in large, continuous compute demand is not just a volume story — it reconfigures the value of energy attributes. Round‑the‑clock, high‑reliability loads increase the premium for dispatchable, long‑duration flexibility (hydro, pumped storage, seasonally firm resources) relative to intermittent generation; in tight regions that premium can translate into sustained basis spreads and capacity payments worth tens to hundreds of $/kW‑yr over a decade. Owners with operating, permitted, or grid‑adjacent sites capture outsized economic rents because build timelines for new transmission and interconnection routinely stretch multi‑year, creating a scarcity wedge.

Supply‑chain and permitting bottlenecks create near‑to‑medium term idiosyncratic winners. Manufacturing lead times for batteries, inverters and large turbines can extend 6–18 months under surge demand, while permitting/transmission reinforcement can take 3–7 years; that time mismatch raises the value of incumbency and long‑term offtakes. Expect contract terms to evolve too — buyers will increasingly pay for shape and firmness (higher fixed capacity/availability fees, longer tenors, indexing clauses), shifting cash‑flow profiles for sellers from merchant volatility toward bankable annuities.

Key reversal risks are clear and quantifiable: a material slowdown in hyperscaler capex, a rapid improvement in chip energy efficiency that reduces kWh per petaflop by >30% in 18 months, or a sustained rate shock that re‑prices asset yields could compress valuation premia quickly. Regulatory intervention (interconnection reform or preferential transmission allocation) would redistribute rents from asset owners toward system operators and could shorten the scarcity window from years to quarters. Monitoring: interconnection queue backlog, announced long‑term PPAs and contractor lead times are high‑signal metrics over the next 6–24 months.

Tactically, allocate to listed owners with large, geographically diversified fleets of dispatchable or storage‑paired renewables and with the balance‑sheet optionality to fund brownfield growth; hedge macro interest‑rate exposure separately. Price in a scenario where contracted cash flows re‑rate by 100–300bps and model total returns over a 12–36 month horizon accordingly; set explicit triggers for de‑risking tied to distribution coverage metrics and announced PPA roll‑outs rather than calendar dates.

Key Decisions for Investors

• Long BEPC (Brookfield Renewable Partners units) — 12–36 month horizon. Position size: tactical overweight (up to 2% NAV). R/R: target 15–25% total return (distributions + modest multiple expansion) vs downside 20–30% on a rate shock or distribution cut. Hedge: short 2‑yr Treasury futures sized to duration sensitivity; de‑risk if distribution coverage falls below 1.1x.
• Pair trade — Long BEPC / Short TAN (Invesco Solar ETF) 1:0.5 — 12 month horizon. Rationale: capture contracted/dispatchable premium vs pure merchant solar exposure should congestion and capacity premiums widen. R/R: asymmetric — limited carry cost, upside if firmness premiums emerge; risk if broad renewable oversupply compresses spark spreads.
• Options carry trade on BEPC — buy Jan 2028 LEAP calls (delta ~0.40) financed by selling 3–6 month covered calls to boost yield. Objective: convex exposure to re‑rating and long‑dated PPA roll‑outs while monetizing near term carry. Risk: time decay and short‑call assignment if short volatility spikes; keep net delta <0.6.
• Long regulated/transmission‑adjacent utilities (e.g., NextEra NEE) — 12–24 months. Rationale: capture grid upgrade tariff recovery and contracted renewables growth with lower merchant exposure. R/R: defend against merchant price swings but sensitive to IRR/utility rate cases; cut position on adverse regulatory rulings or EPS guidance misses.