Warren Buffett's Successor, Greg Abel, Pared Down Bank of America, and Piled Into a Virtual Monopoly That's Now Berkshire's 3rd-Largest Holding

Berkshire Hathaway’s 13F shows a continued 8-quarter reduction of its Bank of America stake, cutting the position by 53% (about 549.5M shares) while shifting capital toward Alphabet after a second consecutive quarter of heavy buying. Berkshire’s AI/tech tilt is reflected in Alphabet purchases tied to a $10B private placement, adding 24.5M Class A and 23.6M Class C shares, lifting Alphabet to a ~$36.6B combined market value and Berkshire’s third-largest holding. The article frames this as supportive of Alphabet’s AI-driven growth (Alphabet Cloud sales up 82% YoY in the June quarter) versus BofA’s heightened sensitivity to Fed rate cuts.

Analysis

This looks less like a generic “Buffett likes X, dislikes Y” read-through and more like a regime shift in how Berkshire is deploying balance-sheet credibility: away from a bank whose economics are now rate-cycle dependent and toward a platform asset with durable pricing power and operating leverage. The second-order signal for financials is that the easiest money in money-center banks may already be behind us; when a long-duration capital allocator starts trimming after a multi-quarter run, it often reflects lower expected future ROE rather than just portfolio hygiene. That is mildly negative for BAC/COF and, by extension, for bank multiples that have been supported by buyback yield rather than organic growth.

Alphabet is the cleaner beneficiary because the market can underwrite its earnings power across multiple cycles, not just the current one. Berkshire’s involvement may matter most in the next 1-3 months as a credibility catalyst: it reduces the “AI is all capex, no payback” discount and can pull incremental allocators into GOOG/GOOGL on any soft pullback. The real second-order winner is not just search, but the whole AI monetization stack: if cloud and model APIs keep converting compute spend into recurring revenue, that supports multiple expansion for the largest hyperscalers relative to software names with weaker distribution.

Contrarian view: the market may overestimate how much one conglomerate’s portfolio tells us about forward fundamentals. The BAC de-risking could be mostly valuation discipline, not a call on bank credit quality, and the Alphabet purchase could be a size/liquidity decision into a fortress balance sheet rather than a declaration that upside is still asymmetric. What would falsify the bearish bank read is a sustained NII inflection or a visible acceleration in buybacks; what would falsify the Alphabet bullish read is AI capex rising faster than cloud margin expansion over the next 2 quarters.

Key Decisions for Investors

• Pair trade: long GOOGL / short BAC for 1-3 months. Thesis is multiple divergence as capital rotates from rate-sensitive balance-sheet beta into durable cash-flow compounding; stop if BAC guides NII stabilization or GOOGL cloud growth decelerates materially.
• Sell BAC strength into any post-13F bounce. The risk/reward favors fading rallies above near-term resistance because the overhang from a persistent large-holder exit can cap multiple expansion until the next earnings reset.
• Add to GOOG/GOOGL on pullbacks rather than chasing. Use any broad Nasdaq weakness as entry; the catalyst path is 1-2 quarters of evidence that AI products are monetizing faster than capex is rising.
• Watch COF and other consumer-finance names as a secondary short list if bank sentiment rolls over. If credit costs stay benign but valuations still compress, the trade is less about losses and more about lower terminal ROE assumptions.
• Set a falsifier on GOOG/GOOGL: if cloud margin fails to improve over the next two earnings prints while capex stays elevated, take profits because the AI premium could re-rate lower.