✦ The Future BetsNarrow moat

The market prices Alphabet as an advertising machine with a cloud attached, and for most of the company's life that was the honest description. It no longer is. Quietly, while everyone argued about whether AI would kill Search, the future portfolio crossed the line from science project to measurable business — and the cheapest part of the stock became the part growing fastest. This page is the inventory of those options, with their gauges attached.

Start with Gemini. The assistant that was an afterthought at 90 million monthly users in October 2024 passed one billion in August 2026 — the fastest any Google product has ever reached that mark¹. A business this size being built inside a company this size, at this speed, has no real precedent even at Google.

Waymo is the second act the market keeps forgetting to price. By late 2025 it was carrying 450,000 paid rides a week and visibly widening its lead over Tesla's robotaxi effort² — essentially all of America's paid driverless miles, earned one fare at a time while rivals still rehearse.

The third option hides inside the cloud line. Google Cloud's backlog has swollen to $514 billion³ — half a trillion dollars of contracted future work — and the TPU beneath it has become the only merchant-scale alternative to Nvidia that an outside buyer can actually order: Anthropic alone contracted for up to one million of them⁴. Behind these sit the moonshots — Willow's quantum milestone, Isomorphic's drug-design engine — real scientific leads that cost billions and earn nothing yet.

Taken together I rate the portfolio narrow: each bet is genuinely strong, and each is genuinely contested — by OpenAI, by Tesla, by Nvidia and the hyperscalers, by physics itself. But narrow on assets this large is a fine thing to get free, and at today's multiple you largely do. The number to watch is Gemini's user curve against ChatGPT's: so long as the gap keeps closing, the future is compounding inside the ad machine's price.