With Nearly $400 Billion in Cash and a New CEO, Is Berkshire Hathaway Stock a Buy?

Berkshire Hathaway ended the quarter with a record cash pile above $397 billion, up from $373 billion in December, while remaining a net seller of stocks with $24 billion sold versus nearly $16 billion bought. The company restarted buybacks in March but only repurchased about $234 million, a modest amount compared with roughly $17 billion in 2022-2023. Operating income rose 18% to $11.3 billion, helped by a 29% jump in insurance underwriting profits, but management signaled no major capital deployment, no breakup plans, and a cautious stance on AI.

Analysis

Berkshire’s cash build is less about caution in the abstract and more about optionality scarcity at this valuation regime. When a mega-cap conglomerate is already trading near its long-run multiple band, incremental buybacks become mathematically unattractive unless management has high conviction that intrinsic value is widening faster than the stock can compound; the tiny repurchase figure implies that hurdle is not being met. That matters because Berkshire’s bid historically provides a floor for sentiment in risk-off tape, but a passive balance-sheet posture removes that support and leaves the shares more exposed to relative-performance drift versus simpler capital-return stories.

The more interesting second-order effect is what Berkshire is implicitly signaling to other capital allocators: the public equity market is not offering enough dislocation to justify large deployment, even with enormous dry powder. That is mildly bearish for broad market breadth, because one of the most patient allocators in the world is effectively saying dispersion remains too tight. It also suggests the next meaningful source of upside may come from underwriting discipline and reinvestment inside operating businesses rather than financial engineering; that favors quality compounders with visible reinvestment runways over balance-sheet stories.

On AI, the “we’ll adopt only when additive” stance is strategically rational but likely underappreciated by investors who are treating AI adoption as a binary productivity lever. Berkshire’s businesses are mostly not the obvious first wave beneficiaries, so the base case is a slow, uneven rollout that creates limited near-term earnings lift while preserving downside if capex is mistimed. In contrast, vendors selling mission-critical infrastructure or workflow software to industrials/insurance may benefit more than the end users themselves over the next 12–24 months.

The contrarian take is that the market may be over-reading inaction as complacency when it may simply reflect a high discount rate on empire-building in a late-cycle tape. If risk assets sell off 10-15%, Berkshire’s relative attractiveness improves quickly because the cash drag becomes a feature rather than a bug. The problem for bulls is timing: absent a broad correction or a materially better acquisition window, the stock may underperform on opportunity cost for several quarters even if the business remains fundamentally sound.

Key Decisions for Investors

• Underweight BRK.B vs. financial-quality peers over the next 3-6 months; the setup is poor for multiple expansion while cash earns little and buybacks are de minimis.
• Pair trade: long AAPL / short BRK.B for 1-2 quarters if you want exposure to capital returns plus AI monetization optionality; AAPL has clearer operating leverage and a more visible buyback cadence.
• Buy 6-12 month downside protection on BRK.B around current levels if available at reasonable premium; the name has less support than usual if market breadth weakens and the cash narrative stops offsetting opportunity cost.
• Rotate incremental capital toward AI infrastructure beneficiaries rather than adopters; prefer NVDA and selected software enablers over industrial end-users for a 12-24 month horizon.
• If BRK.B corrects 10%+ on a market drawdown, start scaling in on the thesis that cash becomes a better asset when dispersion widens; downside is capped by liquidity, upside comes from future deployment optionality.