Berkshire Hathaway closed the first quarter under Greg Abel with $397.4 billion of cash and equivalents, $24 billion higher than year-end. [1] Operating earnings came in at $11.35 billion against a FactSet consensus of $11.56 billion — the first miss under the new chief executive. [2] Net income was $10.1 billion, more than double the year-earlier quarter, with $5.8 billion of investment gains carrying the difference. [3] Tuesday's analyst notes have repriced the company across all three vectors at once.

The line on every desk is Abel's, given on the panel after the Saturday release: "We're not going to do AI for the sake of AI." [4] The CEO told shareholders Berkshire is using artificial intelligence "to solve logical problems in our businesses" — BNSF rail operations, the Geico claims pipeline — but rejected the broader investment thesis that a holding company without an AI capex line item carries an inferior multiple. The framing was repeated by Vice Chairman Ajit Jain, who told the meeting that AI cannot price insurance. The Sherwood News headline that ran Sunday — "Berkshire Hathaway Is The Ultimate Anti-AI Stock" — is now the X-native frame on the position. [5]

Geico operating profit fell 34 percent in the quarter on rate compression and storm losses. [3] Insurance operating profit rose 28 percent across the rest of the segment. The aggregate of those two — and a 9-percent decline in railroad operating profit — is the operating-line miss the desks are pricing. None of the three drags is structural. None of the three is AI-related.

The Apple position remained roughly $80 billion at quarter-end. [3] Berkshire was a net seller of stocks during the quarter for the eighth consecutive quarter. Buybacks ran $234 million, all in March — the first share repurchases since May 2024, and tiny against the $397 billion cash pile. [6] The buyback signal, in the context of Buffett's consistent past framing that Berkshire repurchases only when shares trade below intrinsic value, is the cleanest read on management's view of its own multiple.

The structural frame is the cohort split. Microsoft, Alphabet, Amazon and Meta have committed roughly $710 billion of 2026 AI capital expenditure between them. [7] Apple's calendar-2026 capital plan runs near $13 billion, plus a $100 billion buyback authorization announced last week. Berkshire's $397 billion of cash sits opposite all of that — neither capex nor return, just dry powder. The position is not deferred capital allocation. It is, on Abel's framing, allocation.

The skepticism on the institutional read is whether Abel's first miss puts the operating-line trajectory at risk. The 18-percent year-on-year operating-profit growth Berkshire reported through fiscal-year 2025 has thinned. The Q1 print was the first under a chief executive whose mandate, per the May meeting commentary, is to keep the conglomerate's operating businesses compounding through a period when the holding company is choosing not to bid for the AI infrastructure stack. The pattern of quarterly deceleration that began at year-end 2025 is what next quarter has to break.

The trading day's other signal is Cerebras. The IPO bellwether for the AI-chip category cut its headline from $40 billion to roughly $26.5 billion midpoint Monday, even as indications of interest exceeded $10 billion. [8] The syndicate chose to clear into the Berkshire tape rather than price through it. The two prints — Berkshire's record cash and Cerebras's halved headline — read as the bookend of one referendum on whether the AI capex regime carries a multiple. Tuesday's analyst desks are pricing the answer.

-- THEO KAPLAN, San Francisco