Will Alphabet Break Warren Buffett's Cardinal Rule of Investing?

Alphabet’s ROIC is pressured by AI capex: post-tax ROIC fell from ~58% (2024) to ~42% (2025), with analysts expecting another decline to 38.3% this year and below 31% by 2029. Management guided to roughly $200B capex this year, with CFO stating it is expected to increase significantly in 2027, which widens invested capital and can mechanically reduce ROIC. Offsetting this, Berkshire increased its Alphabet stake dramatically—from ~$5.6B to nearly $37.8B between end-2025 and Q2—highlighting confidence despite near-term return-on-capital headwinds.

Analysis

The key mechanism here is a classic capital-intensity tradeoff: AI spend can preserve strategic position while still impairing near-term equity compounding. For Alphabet, the market should care less about headline growth and more about whether incremental capex is still producing enough operating leverage to justify a premium multiple; if not, the stock can underperform even while the business remains high quality. The clearest second-order winner is the AI infrastructure stack, where hyperscaler spending becomes revenue visibility for chip, networking, and power-adjacent suppliers rather than for the buyers of that capacity.

Berkshire’s buying is better read as a valuation/quality screen than as proof that the ROIC trend is benign. The real catalyst path is the next 1-3 earnings prints and 2027 guidance: if management keeps signaling higher spend before monetization is visible, the market will start treating the stock as a capital allocator story, not a compounding story, and that usually triggers multiple compression over several quarters. The thesis is falsified if AI products begin contributing measurable revenue acceleration or if capex flattens earlier than expected, allowing free cash flow conversion to inflect.

Contrarian view: consensus may be too quick to equate falling ROIC with broken economics. Alphabet can likely sustain returns above its cost of capital for years, and that makes a full bearish call on the franchise premature. The better edge is relative value: prefer exposure to the sellers of AI infrastructure over the buyers, while treating Berkshire’s position as supportive but not decisive.

Key Decisions for Investors

• Long NVDA / short GOOGL into the next 1-2 earnings cycles: express the view that AI capex accrues more to infrastructure suppliers than to hyperscaler buyers. Risk/reward is favorable if Alphabet keeps pushing capex higher, but cover quickly if monetization commentary improves materially.
• If already long GOOGL, hedge with a 3-6 month put spread sized against the next capex-guidance update rather than exiting outright. This limits damage from another step-up in spending while preserving upside if AI monetization surprises higher.
• Use BRK.B only as a sentiment monitor, not a standalone buy signal. Berkshire’s Alphabet stake does not materially change BRK.B earnings power, so there is no immediate catalyst for a fresh BRK.B trade unless position disclosure or capital allocation commentary changes.
• Add to a GOOGL short only on confirmation: another quarter of rising capex with no offsetting cloud/search acceleration. The trade works on repeated multiple compression, not on one data point.