Abel: Two ways Berkshire hopes to cash in on AI

Berkshire CEO Greg Abel said the AI data-center buildout is a “significant opportunity” for Berkshire Hathaway Energy, but any power sales to hyperscalers would require “no impact” to other customers’ rates; he cited Iowa data centers contributing about 8% of utility load and generating “very, very substantial” local tax relief. Abel also outlined an AI-linked investment pathway via Alphabet, referencing Berkshire’s ~$10B purchase of Alphabet shares at a ~6.5% discount as Google raised ~$80B for AI compute infrastructure. On Japan rates, he noted the 10-year yield just above 3% is still “relatively modest,” and Berkshire’s yen bond portfolio has a remaining life of a little over five years, supporting continued debt issuance for carry as dividends and buybacks are expected to rise.

Analysis

The investable signal here is not “AI is good”; it is that the scarce resource in the buildout is shifting from GPUs to power, permits, and community acceptance. That favors regulated utilities and vertically integrated infrastructure owners with rate-case credibility, because they can monetize load growth without taking technology risk. The flip side is that the bottleneck is political, so the market may be underestimating how often projects get delayed by local opposition, water constraints, or customer-rate scrutiny.

Alphabet’s financing signal matters more than the headline stake size: a discounted primary suggests management still sees a high-return internal reinvestment runway rather than a mature business milking cash. That supports the bull case for GOOGL, but it also implies elevated capex pressure can cap near-term margin expansion; the stock should trade on proof of monetization, not AI narrative alone. Over 1-3 months, any disappointment in cloud growth or operating leverage would likely hit the multiple faster than Berkshire’s endorsement can lift it.

Japan looks like a carry story that is changing slowly, not breaking. Modestly higher JGB yields should not impair the trading houses’ economics, but a stronger yen would be the real P&L risk for USD investors in the basket. Housing remains the weakest leg: the market may be too quick to price in a smooth rate-cut recovery, while the data still point to a drawn-out demand normalization rather than a sharp rebound.

Key Decisions for Investors

• Go modest long GOOGL vs. QQQ into the next earnings window; use a 3-6 month horizon and require cloud growth plus capex discipline to confirm the AI thesis. Falsify if operating margin or FCF guidance is cut again.
• Add BRK.B on pullbacks as a medium-duration infrastructure compounder, not a catalyst trade; the underappreciated upside is utility load growth tied to AI. Exit or trim if regulators block data-center load additions or force uneconomic rate concessions.
• Stay short/underweight LEN or XHB for 1-3 months; this is a timing trade against consensus hoping for a quick housing turn. Cover if mortgage rates fall meaningfully or new-order trends re-accelerate.
• Own the Japanese trading-house basket only with explicit FX hedging (ITOCY, MITSY, MTSUY, SSUMY, MARUY); the main risk is yen strength, not domestic rates. Falsify if USD/JPY breaks sharply lower and local-currency earnings do not offset it.
• No trade on Tokio Marine rumor until there is a signed transaction; treat it as an alert, not an edge.