The company that indexed the web is now betting everything on an AI it calls Gemini - and for the first time, the search box has competition it built itself.
Type a question into a box, and within a fraction of a second the sum of the internet reorders itself to answer you. Most people do this dozens of times a day without a second thought. That reflex, repeated across billions of people, is the foundation of one of the largest businesses in human history. It began in 1996 as a graduate-school experiment at Stanford, where two students named Larry Page and Sergey Brin were curious about a simple question: what if you ranked web pages by how many other pages linked to them?
They called the project BackRub. It was renamed Google - a play on "googol," the number one followed by a hundred zeros - and incorporated on September 4, 1998. The early legend is now Silicon Valley scripture: Sun Microsystems co-founder Andy Bechtolsheim watched a demo in a driveway and wrote a $100,000 check to "Google Inc." before the company legally existed. The first office was a rented garage in Menlo Park.
Nearly three decades later, the garage has become a subsidiary of Alphabet Inc. that reaches more people, more often, than almost any product ever made. And the reflex it built its fortune on - type, click, answer - is the very thing it is now trying to rebuild.
At its core, Google organizes information and puts it one click away. The mission statement written in 1998 - "to organize the world's information and make it universally accessible and useful" - has never been retired, and it still explains the sprawl. Search was the first expression of it. Then came the products that most people now treat as infrastructure: Gmail for email, Maps for getting around, Chrome for browsing, Android for phones, YouTube for video, Google Photos for memories, Workspace for getting work done.
Six of these products each have more than a billion users. That is the number worth sitting with. Not a billion downloads or a billion sign-ups - a billion people who reach for it as a habit.
There is a tidy way to understand Google that cuts through the product list: the people who use it are not the people who pay for it. Billions of consumers use Search, YouTube, Maps and the rest for free. The paying customers are advertisers - businesses of every size that want a slice of all that attention and intent. When you search "running shoes," you are not the buyer. You are the inventory.
Google gives away the software and sells the attention. It is the oldest trick in media, executed at a scale nobody has matched.The advertising model, in one line
That is one audience. The other is enterprise. Companies, startups and governments buy Google Cloud for computing power, Workspace for productivity, and increasingly the Gemini models for AI. Schools run on Chromebooks and Classroom. Developers build on Google's APIs. This second audience pays directly, and it is growing faster than the first.
In 2025, Alphabet's annual revenue passed $400 billion for the first time. The bulk of it still comes from advertising - Search ads, YouTube ads, the display network - but the shape of the business is shifting. Google Cloud grew 48% to an annual run rate of roughly $70 billion. YouTube, counted quietly inside "Services," did more than $60 billion across ads and subscriptions. If YouTube were spun out, it would rank among the largest media companies on earth.
Operating margins run above 30%. The reason a free search box can fund self-driving cars, quantum computing and a dozen moonshots is that intent-based advertising is one of the most profitable machines ever built. Someone searching to buy is worth a great deal to someone selling.
For 25 years, Google solved a single durable problem: the internet is vast and disorganized, and people need to find the one useful thing inside it. Ten blue links, ranked well, answered that. But the problem is quietly changing. People increasingly want the answer itself, not a list of places that might contain it - and they want software that can act on their behalf, not just point the way.
That is the pivot behind Gemini, Google's family of AI models launched in 2023. At Google I/O in 2026, CEO Sundar Pichai opened by declaring the company was "firmly in our agentic Gemini era" - agents that book, draft, plan and execute rather than merely retrieve. The Gemini app crossed 900 million monthly users, more than double the year before. Search itself is being rebuilt around AI answers.
Plenty of companies have good AI models. Fewer own the whole stack beneath them. Pichai's argument for why Google is different is vertical: it designs its own TPU chips at the bottom, trains the Gemini models in the middle, runs Google Cloud and Workspace on top, and ships all of it through a distribution network of billion-user products. When he says the company is "compute constrained," it is because demand for that stack outruns supply.
Sundar Pichai's argument is that owning the chips, the models, the cloud and the distribution at once is a moat rivals can copy in pieces but not as a whole.The vertical-integration case
The proof point arrived in January 2026, when Apple agreed to run future Siri and Apple Intelligence features on Gemini. Alphabet touched a $4 trillion market capitalization the same month. When your fiercest hardware rival licenses your AI, the "Google is behind" narrative gets harder to hold.
Google does not have one competitor; it has a different one in every room. Microsoft comes at search, cloud and productivity with Bing, Azure and Copilot. Amazon contests cloud with AWS and now advertising too. Meta fights for ad dollars and AI attention. Apple presses on privacy and controls the iPhone. And OpenAI and Anthropic pushed the frontier of AI assistants hard enough that, for a stretch, they looked capable of doing to Google Search what Google once did to the phone book.
The most interesting rivalry, though, is internal. Every AI answer that satisfies a user without a click is a small threat to the ad-supported link. Google is, deliberately, building the thing that could erode its own core - on the bet that if search is going to be reinvented, better to be the one holding the shovel.
In the market's mind, Google occupies three positions at once. It is a utility - the plumbing so many people route their digital lives through that "google it" became a verb. It is an advertising empire, the largest of its kind. And it is a live question mark: can a company that perfected one way of finding information successfully replace it with another, without losing the machine that made the whole thing possible?
The founders warned about this in their 2004 IPO letter. "Google is not a conventional company," they wrote. "We do not intend to become one." Two decades and $400 billion in yearly revenue later, the line reads less like modesty and more like a standing dare. The reflex it built - type, click, answer - is being rebuilt in real time. Whether the second version prints money like the first is the story worth watching.