Berkshire Hathaway Has a New Top Holding, and It Dominates Its Market

Berkshire Hathaway increased its Alphabet stake by 45% in Q2 to roughly 79 million shares worth $28.2 billion, making Google’s parent 9.41% of its portfolio and its fourth-largest holding. Alphabet delivered Q2 revenue growth of 24% to $120 billion and operating-income growth of 30% to about $41 billion, but raised 2026 AI-infrastructure capex guidance to $195 billion-$205 billion. The investment drove a $5.8 billion quarterly free-cash-flow outflow, Alphabet’s first negative FCF quarter as a public company, contributing to a 7% one-month share-price decline despite a 37% gain over the past year and a valuation of about 16x earnings.

Analysis

The investable issue is not Berkshire’s ownership but whether Alphabet can convert infrastructure spend into incremental Cloud and AI revenue faster than depreciation, power, and networking costs reset its margin structure. If utilization ramps as implied, the market can re-rate GOOG from a "capex casualty" toward a capacity-constrained AI beneficiary; if utilization lags, the stock’s apparently low earnings multiple is misleading because consensus EBIT and FCF estimates will need to fall. The reported investment and free-cash-flow figures should be reconciled to Alphabet’s 10-Q and Berkshire’s filed 13F before treating either as a positioning signal.

Near term, Berkshire-related flow is largely stale and should not create durable technical support; the more relevant 1-3 month catalyst is management’s next capex-to-revenue conversion evidence: Cloud growth, remaining performance obligations, depreciation guidance, and buyback pace. A sustained deterioration in FCF would pressure not only GOOG but also AI-infrastructure suppliers with revenue concentration in hyperscalers, particularly NVDA and networking/power beneficiaries, because it would signal a lower return threshold for capacity expansion. Conversely, a credible demand backlog with stable Cloud margins would validate the spend and likely compress the valuation gap versus AMZN and MSFT.

The contrarian opportunity is that investors may be treating all hyperscaler capex as equivalent. Alphabet’s search cash engine can subsidize a longer buildout, but that advantage only matters if AI answers do not cannibalize higher-margin query economics faster than monetization improves. The key structural risk over 6-18 months is therefore not capex alone; it is a simultaneous rise in inference costs and decline in paid-click economics, which would make incremental AI revenue materially less profitable than the legacy business.

Key Decisions for Investors

• Maintain a conditional long GOOG/GOOGL watch position rather than chase the recent pullback; add only after the next earnings release shows Cloud growth acceleration and stable or improving Cloud operating margin. Upside is a multiple re-rating on validated capacity utilization; exit if capex rises again while consensus 12-month FCF falls by more than 10%.
• Express relative-value exposure via long GOOG / short AMZN only if GOOG trades at a meaningful forward-EV/EBIT discount while both companies report comparable cloud-demand strength. Use a 3-6 month horizon; the thesis fails if AWS re-accelerates materially faster or Alphabet’s search monetization decelerates.
• Treat NVDA as a read-through hedge, not a direct beneficiary of the headline: reduce incremental long exposure if the next hyperscaler reporting cycle shows capex guidance cuts or weaker AI backlog disclosures. A broad spending retrenchment would likely hit supplier estimates before it is fully reflected in GOOG earnings.
• Set an event alert for Alphabet’s next capex, depreciation, and repurchase guidance. A capex increase paired with buyback restraint and no corresponding Cloud backlog disclosure is a bearish confirmation; a stable capex envelope with improving FCF conversion is the signal to scale the GOOG long.