Greg Abel Just Quantified Berkshire's AI Power Opportunity

Berkshire Hathaway CEO Greg Abel sees a clear AI-related opportunity, primarily through Berkshire Hathaway Energy (BHE) supplying power for data centers. Abel said data centers accounted for 8% of MidAmerican Energy’s demand in Iowa, while BHE has 32,400 net MW of generation capacity in operation/under construction and plans ~$33.5B of capex from 2024–2026 to expand generation, storage, and transmission. Berkshire also increased its Alphabet stake this year, positioning the conglomerate to benefit from AI both via equities and utilities exposure.

Analysis

The investable takeaway is not “Berkshire owns more AI,” but that AI load is becoming a grid and dispatch problem before it becomes an equity-selection problem. That favors owners of scarce, regulated, long-lived infrastructure—where incremental demand can compound through rate base growth for years—more than it favors the model builders themselves. For BRK.B, the market may underappreciate that BHE can turn data-center load into durable earnings power, but the payoff is slow because utility economics are gated by approvals, interconnection, and construction cycles.

Second-order winners are the firms that can add firm power quickly: transmission, gas-fired generation, and power equipment. The losers are the hyperscalers if power becomes the bottleneck on deployment cadence, plus oil-first exposures that do not directly monetize electrons; this is more bullish grid and gas than crude. GOOGL benefits from Berkshire validation, but that is mostly sentiment and quality signaling unless it coincides with better AI monetization or slower capex intensity.

The contrarian miss is that the market may overread “AI exposure” as a beta trade when the real constraint is local capacity and regulatory pass-through. If regulators force BHE to socialize costs without acceptable returns, or if data-center demand slips, the thesis weakens quickly. The key falsifiers over the next 1-3 months are utility commission outcomes, interconnection queues, and whether hyperscalers keep announcing load without corresponding power commitments; over 6-18 months, watch rate-base growth and authorized ROE rather than headline AI adoption.

Key Decisions for Investors

• Long BRK.B on pullbacks for a 6-18 month horizon; thesis is steady uplift from regulated load growth, but cap upside is modest unless BHE converts data-center demand into above-average rate-base expansion.
• Initiate a basket long in firm-power beneficiaries vs XLU (e.g., CEG/VST/ETN on one side, short XLU on the other) for 1-3 months; this is a better expression of the AI-power bottleneck than owning pure oil beta.
• Use this as a watchlist trigger for GOOGL rather than a chase: add on weakness only if AI monetization metrics improve and capex intensity does not outrun operating leverage.
• Do not buy CVX/OXY purely on this headline; if already long energy, trim into strength and rotate toward gas/power names that monetize incremental electricity demand more directly.