Global Capital

Berkshire's AI Bet

Berkshire — the board's largest true dry powder and famously the tech-skeptic conglomerate — built a fast-growing Alphabet position ($4.3B disclosed Q3'25 → ~$28-41B incl. a $10B AI-infra private placement Jun'26) that Buffett himself frames as an AI-capex bet, even as new CEO Abel says "we're not going to do AI for the sake of AI." Track: the Q2 2026 13F for whether the position kept growing or any other AI-adjacent name appears, and further Buffett/Abel commentary. Secondary, lower-priority strand: GEICO's June 2026 PA AI-underwriting settlement and a reported (unconfirmed-subsidiary) AI-liability policy exclusion — industry pattern more than Berkshire-specific.

STATUS · OPEN OPENED · 2026-08-04 LAST SEEN · 2026-08-04

berkshire-hathaway

2026-08-04 — Thread opened + crawled ⟨crawl 2026-08-04⟩

Opened following a board-pass audit that found the entire insurance pocket with zero live thread coverage. Berkshire’s own board entry already tracks its general M&A/capital behavior; this thread is specifically the AI-adjacent capital-allocation and investing-stance story that entry doesn’t cover.
The Alphabet position built rapidly across the last year: first disclosed Q3 2025 (17.8M shares, ~$4.3B) -> ~54M shares by Q1 2026 -> a $10B private placement on 2026-06-01 when Alphabet raised $80B in new equity specifically to fund AI infrastructure (split $5B Class A / $5B Class C) — combined position estimated $28-41B depending on the article’s date. ⟨crawl 2026-08-04⟩
Buffett, on CNBC 2026-07-15, said he personally initiated the position and tied it directly to the AI spending race: “The real question with Google and all of its competitors now, because they’re all laying out hundreds of billions, and that’s real money.” He called the buildout “less attractive” than other Berkshire holdings and said he regretted not buying in earlier — pragmatic acknowledgment of the AI capex race, not tech enthusiasm. ⟨crawl 2026-08-04⟩
Abel, at his first annual meeting as CEO (2026-05-02), drew the opposite institutional line: “We’re not going to do AI for the sake of AI” — AI gets adopted only where it “clearly benefits” Berkshire’s core businesses. The meeting opened with a staged AI deepfake of Buffett himself, deliberately used to spotlight deepfake/AI-security risk; Abel: “It’s a great reminder for our team… That is a significant risk across Berkshire that we’re managing every day.” Buffett separately called deepfake risk “scary,” citing specifically a convincing deepfake of a nuclear-armed head of state. ⟨crawl 2026-08-04⟩
Apple + Alphabet are now ~30% of Berkshire’s ~$351B equity portfolio (2026-07-14), after Abel cut the book from 42 to 29 holdings in Q1 2026 alone. No Nvidia or other pure-play AI-infrastructure name appears in the Q1 2026 13F; Q2’s wasn’t yet public as of this crawl. ⟨crawl 2026-08-04⟩
Secondary strand — insurance-side AI, industry pattern more than a Berkshire-specific story: Pennsylvania’s AG settled with GEICO in June 2026 over an AI underwriting-review tool that flagged a new policyholder for extra documentation without adequate notice, leaving her unknowingly uninsured (GEICO added a week to the response window, committed to NAIC AI Model Bulletin governance, settled without admitting wrongdoing). Separately, Berkshire is reported (with Chubb, Travelers, AIG) to have won state approval to add AI-liability exclusions to commercial policies it underwrites — the specific subsidiary and exact date remain paywalled/unconfirmed. ⟨crawl 2026-08-04⟩

The Projection

Berkshire's AI Bet

2026-08-04 — Thread opened + crawled ⟨crawl 2026-08-04⟩

← Backstory