Prediction: Following Alphabet, This Could Be the Next Trillion-Dollar Artificial Intelligence (AI) Stock Greg Abel Adds to Berkshire's Portfolio

Berkshire Hathaway initiated a new Alphabet position in Q3 2025 and then more than tripled it to nearly 54 million shares in Greg Abel’s first full quarter as CEO, while also fully exiting Amazon. The article argues Nvidia could fit Berkshire’s evolving tech posture, citing Nvidia’s CUDA software moat, expansion across the AI stack (networking, optical components, and custom-chip enablement), and strong profitability with shares up over 1,000% in recent years. It also notes Nvidia’s forward P/E is closer to reasonable relative to growth, implying less expensive than earlier AI-cycle phases.

Analysis

The actionable signal is not “Berkshire may buy NVDA”; it is that the market is increasingly rewarding AI exposure that looks like a durable toll road rather than a one-cycle hardware spike. That favors the names with embedded software/networking lock-in and recurring platform pull-through, while leaving the more interchangeable semiconductor and neocloud adjacencies vulnerable if capex growth normalizes. In that framework, NVDA remains the cleanest way to express AI infrastructure durability; MRVL/COHR/LITE/CRWV can move harder on flow, but their business models are more exposed to spending pauses and margin compression.

The second-order loser is AMD, not because this changes near-term fundamentals, but because it reinforces the “best-in-class wins the wallet share” narrative in AI accelerators. AMZN is more of a sentiment casualty than a fundamental one: if Berkshire’s posture is read as favoring simpler, higher-visibility compounding stories, the market may apply a relative valuation penalty to asset-heavy platform names whose AI payoff is harder to isolate. GOOGL benefits from the same quality-growth framework, and BRK.B may get a modest multiple tailwind if investors extrapolate Abel as more willing to own secular compounders.

The real catalyst path is the next 1-3 months: NVDA earnings, hyperscaler capex commentary, and any Berkshire filing/annual-letter signal. If order growth or gross margin inflects down, the “AI compounder” re-rate can reverse quickly because this trade is currently supported more by narrative and positioning than by new fundamental evidence. Over 6-18 months, the key falsifier is a shift from general-purpose GPUs toward custom ASICs or a visible slowdown in cloud build-outs; that would compress the ecosystem multiple even if unit demand stays healthy.

Key Decisions for Investors

• Long NVDA / short AMD for 1-3 months: express best-in-class moat versus a credible but weaker share-taker; risk/reward is favorable if NVDA keeps pricing power and AMD remains a secondary beneficiary, but stop if AMD closes the gap on inference wins or NVDA gross margin misses.
• Buy GOOGL on weakness rather than chase NVDA here: Berkshire-style quality-growth exposure with less capex intensity and a cheaper base multiple; better 6-18 month risk/reward if ad/search remains stable and AI monetization continues to scale.
• Treat COHR/LITE/MRVL/CRWV as trading vehicles, not core longs: consider small tactical longs only into pullbacks ahead of NVDA capex commentary; these names can outperform on flow, but they are the first to de-rate if AI spending pauses.
• Avoid adding to AMZN on the Berkshire narrative alone; if anything, use strength to fade relative performance versus GOOGL or NVDA because the article highlights a preference for simpler compounding stories, not a deterioration in Amazon’s operations.
• Set an alert for the next Berkshire 13F and NVDA earnings: if no incremental Berkshire confirmation appears and NVDA guidance is merely in-line, fade any post-article multiple expansion by 5-10% as the move is likely narrative-driven rather than fundamental.