How to Invest $267 Billion? Buffett’s Final 2025 Q3 Portfolio Holds the Value Code for the AI Era

As Q3 2025 draws to a close, the market has obtained the final portfolio report from Warren Buffett during his 60-year tenure as CEO of Berkshire Hathaway — this report carries seismic signaling weight: Berkshire executed three key moves in Q3: making its first purchase of approximately $4.3 billion in Google stock, trimming its largest holding Apple (a 15% reduction from the prior quarter), and boosting its cash reserves to a record high of $381.7 billion.

Amid the global AI investment frenzy and debates over AI bubbles, what is the 95-year-old Buffett planning?

First, Understand Buffett: The “Don’t Invest in What You Don’t Understand” Rule — It Helped Him Dodge Bubbles, But Also Miss Opportunities

Over his 60-year investing career, Buffett’s core principle has been his “circle of competence”: he steadfastly avoids bubbles he cannot grasp.

The most iconic example is the 2000 dot-com bubble — while the world went crazy for the internet, he bought no shares, reasoning, “I don’t know what these companies will look like in 10 years, but I know Coca-Cola will still be selling soda.”This rule helped him evade bubbles, but also caused him to miss early opportunities with Amazon and Google; later, he and Charlie Munger publicly admitted missing Google was a mistake, given that GEICO (Berkshire’s insurance subsidiary) was one of Google’s major advertisers, and they should have understood its value.

The only tech stock in his later years was Apple — but his logic was: “Apple isn’t a tech stock; it’s a consumer goods company with an ecosystem moat.”

Why Google? The $4.3B Buy Is a “Mispriced AI Infrastructure Play”

Amid the AI boom, Buffett didn’t chase NVIDIA — instead, he chose Google, driven by two core logics: “margin of safety + predictable cash flow”:

Undervalued valuation: Market fears that “OpenAI would disrupt Google’s search business” depressed its valuation — Alphabet (Google’s parent company) had a P/E ratio of just 25x, far lower than the 30+x multiples of Microsoft and NVIDIA, perfectly aligning with his “buy a dollar’s worth of assets for 40 cents” margin of safety principle.

Stable business fundamentals: Alphabet’s Q3 2025 revenue topped $100 billion in a single quarter, with its core search and YouTube businesses remaining strong; Google Cloud generated $15.2 billion in revenue, up 34% year-over-year, rapidly capturing market share.

Deep AI moat: The AI race has three layers — computing power, platform, and models — and Google is one of the few companies heavily invested in all three: its Gemini model can compete head-to-head with GPT, and the high computing costs of AI search have raised industry barriers; he avoided NVIDIA because he “dislikes suppliers dependent on a few big clients” — NVIDIA relies heavily on cloud giants like Microsoft and Google, all of which are developing their own AI chips.

Why Trim Apple? From 40% to 23% — Risk Diversification and a Signal of the Times

When Buffett chased Apple in 2016, he was drawn to its “user stickiness + ecosystem moat”; but in 2025, Apple faces dual challenges:

· Regulatory pressure looms: Global regulators are targeting Apple’s 30% App Store commission — the EU’s Digital Markets Act, UK court rulings, and U.S. Department of Justice antitrust lawsuits are all undermining its core profit logic.

· Slow AI progress: Apple has been relatively cautious in the large language model race, even nearing a deal to pay Google $1 billion annually for a custom Gemini model to upgrade Siri.

More critically, this move is about risk diversification: Apple once accounted for over 40% of Berkshire’s portfolio, creating excessive single-position risk; the cut to ~23% is a proactive adjustment to reduce concentration.

$381.7B in Cash: Not Conservatism, But “Offensive Defense”

Berkshire is fundamentally an insurance company, investing with “float” (free, long-term capital), and the $381.7B cash pile is a “dual-purpose offensive-defensive” layout:

Passive interest income: In the current ~5% interest rate environment, this cash generates ~$20 billion in annual interest income, becoming a “new opportunity cost.”

Waiting for market panic: The massive cash reserve is “purchasing power” prepared for potential future market corrections — Buffett’s classic playbook is “be greedy when others are fearful,” and this cash is his ammunition.

Paving the way for succession: At 95, he is set to step down as CEO; buying Google balances “Old Buffett’s value principles” with the new team’s tech vision and the incoming CEO’s preference for infrastructure.

Conclusion: Buffett’s Era Shift — From Consumer Brands to Mobile Internet to AI Infrastructure

Over 60 years, the “era moat” in Buffett’s eyes has evolved:

1980s (consumer brand era): The moat was Coca-Cola’s brand;

2010s (mobile internet era): The moat was Apple’s ecosystem;

2020s (AI era): The moat is AI infrastructure like Google.

His value investing has never rejected technology — instead, it redefines “predictable long-term cash flow”: Google’s 20 years of search data and 34% growth in cloud services make it the “understandable certainty” of the AI era.

As his logic goes: “I never invest in the stock market — I invest in the cash flow of an era”; true long-termism isn’t clinging to a single stock, but upholding a set of principles — he never bets on companies, but on the “profitable river” of the era.