Warren Buffett Just Confirmed That Berkshire's Alphabet Stake Was His Idea. Why the Stock Still Looks Like a Buy

The incremental signal here is not “Buffett likes a tech name,” it’s that the market now has a quality-growth framework for valuing Alphabet’s AI spend: if the company can self-fund compute, monetize it through ads/cloud, and keep unit costs below peers, the capex story becomes margin-accretive rather than dilutive. That matters most for GOOG/GOOGL multiple durability, while AVGO gets a secondary boost as the TPU supply chain becomes a higher-confidence revenue stream. NVDA is the relative loser because any credible custom-silicon adoption narrative increases the odds of share loss at the margin and keeps pressure on its premium multiple.

Near term, the move is mostly sentiment and can fade after the first reaction; the real catalyst path is 1-3 months of cloud commentary, TPU ordering evidence, and whether AI spend shows up in operating margin instead of just capex. The thesis breaks if Alphabet’s infrastructure buildout outpaces monetization, cloud margins stall, or management signals that custom silicon cannot keep up with model demand. Over 6-18 months, the structural question is whether Alphabet can use lower-cost inference to defend search economics while taking cloud share without sacrificing ROIC.

The consensus may be missing that this is as much a competitive-cost story as an AI-growth story: cheaper compute can expand Alphabet’s pricing flexibility and pressure rivals to either accept lower returns or increase spend. That makes this a better relative-value long GOOG than a blanket long-NVDA call. If the market starts treating Alphabet as a high-quality compounder with AI optionality rather than a pure ad company, downside should compress faster than upside, but only if execution remains clean.