Greg Abel Says Berkshire Will Serve the Hyperscalers Only If Its Other Customers Don't Pay for It

Berkshire Hathaway CEO Greg Abel says its AI hyperscaler data-center power deals must deliver a “net benefit” with no rate impact to other utility customers. In the latest results, after-tax earnings at U.S. utilities rose 38% YoY to $597M in Q2, supported by higher retail volumes (up ~3% YoY through June) and an 8% YoY increase in electric utility margins. Berkshire is investing heavily—$6.7B of $10.6B first-half capex went to energy (plus BNSF), with another ~$8.6B expected in 2H 2026—despite growing community pushback that has not yet blocked projects.

Analysis

This is less a “AI demand is good” story than a screening mechanism for who gets to monetize load without triggering a political or regulatory tax. A utility that can prove hyperscaler demand is incremental to existing customers effectively turns the buildout into durable rate base growth, which is why BRK.B should compound faster than utilities that have to fight every incremental MW in a rate case. The strategic edge is not just load growth; it is the ability to approve it with lower probability of socialized-cost backlash.

The second-order winners are the grid spend and firm-power ecosystem, not the data-center operators themselves. The biggest earnings leverage sits in transmission, substations, transformers, and construction, which points more toward GEV, ETN, and PWR than toward the hyperscalers; over 6-18 months, these are the names where orders can outrun sentiment. For BRK.B, the immediate earnings delta is limited because capex precedes allowed returns, so the stock should be viewed as a multi-quarter compounding story rather than a near-term EPS catalyst.

The contrarian risk is that the market may be extrapolating data-center load too linearly while underestimating how restrictive Abel’s “net benefit” شرط really is. That condition should make approvals more durable, but it also means fewer projects clear the bar than bulls assume, and some states will still slow or cap deals. The thesis breaks if states push explicit cross-subsidy rules, if a major community rejection stalls a flagship site, or if behind-the-meter generation reduces utility capture of the capex cycle.

Key Decisions for Investors

• Long BRK.B vs. XLU on a 3-6 month horizon: BRK’s utility growth is more defensible and less politically fragile; target relative outperformance if the market starts rewarding regulated growth visibility. Thesis fails if rate cases across the sector re-rate utilities broadly and XLU rallies on falling rates.
• Add GEV and ETN on pullbacks over the next 1-3 months: these are the cleaner second-order beneficiaries from grid interconnect and backup-power capex. Risk/reward is better than chasing hyperscaler equities because the revenue is tied to physical bottlenecks, not sentiment.
• Pair trade idea: long BRK.B / short AEP for a 3-6 month regulatory-risk hedge if data-center headlines intensify in contested states. BRK’s approval model is more community-friendly; the short works only if AEP’s growth story runs into siting or cost-allocation friction.
• If you want a slower, lower-beta expression, buy BRK.B on any post-earnings pullback and hold 6-18 months. The upside is steady rate-base compounding; the thesis is invalidated if utility earnings growth decelerates sharply or if management signals a lower-than-expected capital deployment runway.