Google Gemini is winning the part of the AI race that gets headlines and losing the part that pays the bills. The models are strong, adoption is enormous, and Alphabet's stock reflects a company that looks anything but disrupted. Yet the same technology that makes Gemini impressive also chips away at the ten-blue-links habit that turned Google into one of the most profitable businesses ever built. That is the tension worth understanding, and it is more interesting than another benchmark chart.

Alphabet crossed $400 billion in annual revenue for the first time in 2025, finishing the year at roughly $402.8 billion. Advertising is still the engine: Google's ad business pulled in about $81.5 billion in the fourth quarter alone, up 14% year over year. Search specifically grew faster than it had in years. So the crisis narrative around AI eating Search has, so far, not shown up in the numbers. The question is whether that holds as behavior shifts.

How DeepMind turned Google Gemini into a real contender

For a while, Google looked flat-footed. ChatGPT launched, the internet declared Search dead, and Google's first public demos were rocky. The turnaround came from consolidating its AI research under Google DeepMind and shipping the Gemini family as thinking models, systems that reason through a problem before answering rather than predicting the next token in one pass.

The Gemini 2.5 line, introduced at Google I/O 2025, is where this became obvious. Gemini 2.5 Pro posted leading scores on hard reasoning, math, and coding benchmarks, and a mode called Deep Think lets the model weigh multiple hypotheses before committing. Whether or not you trust any single benchmark, the direction is clear: Google closed most of the capability gap that briefly made it look vulnerable.

Distribution is the part rivals cannot easily copy. The Gemini app surpassed 750 million monthly active users by Alphabet's Q4 2025 earnings. Google also has Android, Chrome, Workspace, and a Search product that touches a large share of the planet's internet users. A model is a feature. A model wired into products billions of people already open every day is a moat.

Why AI answers threaten the business that funds them

Here is the awkward math. Google makes money when you search, scan a page of results, and click something, ideally an ad. Every one of those clicks is an auction. An AI answer that resolves your question at the top of the page removes the reason to click at all.

This is the classic innovator's dilemma, and Google is living it in public. The company cannot ignore conversational AI, because if it does, users drift to assistants that answer directly. But leaning into AI answers risks compressing the very ad inventory that generates most of Alphabet's profit. Every improvement to the answer is a small subtraction from the funnel.

Publishers feel this first. When a summary satisfies the query, the sites that supplied the underlying information get fewer visits, which hits their own ad and subscription revenue. That is not a side effect; it is the mechanism. The open web that trained these models depends on traffic the models are learning to intercept. If that loop tightens, the quality of freely available information that feeds future models degrades, a slow problem that will not show up in a quarterly report.

Google Gemini inside Search: cannibalize yourself first

Google's answer is to run the disruption internally rather than let a competitor run it. AI Overviews, the Gemini-powered summaries that sit above traditional results, now reach on the order of 2 billion users a month. AI Mode goes further, turning Search into a chatbot-style conversation with follow-up questions.

The commercial fix is to put ads inside the AI experience. Google has confirmed it is placing Search and Shopping ads within AI Overviews and AI Mode, marked as Sponsored. The bet is that a well-placed ad inside a relevant answer can monetize as well as, or better than, a link in a list, because intent is clearer and the surface is richer.

Early evidence from advertisers is mixed and worth reading skeptically, since much of it comes from agencies with something to sell. Research- and education-stage queries, common in B2B and SaaS, appear to lose the most, because those are exactly the questions an AI summary answers outright. High-intent commercial and local searches look steadier. The through-line: the top of the funnel is being absorbed by answers, while the bottom of the funnel, where people are ready to buy, still rewards a paid placement.

What operators should actually watch

Stop tracking whether Google Gemini beats a rival on a leaderboard. It is competitive, and rankings churn monthly. Watch these instead.

First, ad revenue per query. If AI answers were destroying the business, you would see ad growth stall even as usage climbs. Through 2025 it did the opposite, which tells you monetization is keeping pace with the format shift, for now.

Second, referral traffic from Search to independent sites. This is the leading indicator of a compressed open web. If it falls sharply, expect publisher business models, and eventually content supply, to strain.

Third, cloud. Google Cloud grew 48% in Q4 2025 to $17.7 billion, and much of the Gemini story is about selling the same models to enterprises through the cloud. That is a cleaner, faster-growing revenue line with none of the cannibalization problem, and it is where Google would rather you look.

The honest read: Google Gemini has neutralized the existential threat that pundits declared in 2023, and Alphabet is managing the transition from a position of strength. The unresolved question is structural. A company optimized to sell clicks is teaching the world to expect answers. It has the distribution and the balance sheet to run that experiment on itself. Whether the answer economy pays as well as the click economy is a bet still being placed, one quarter at a time.

Frequently Asked Questions

Is Google Gemini losing to ChatGPT?

Not on distribution. The Gemini app passed 750 million monthly active users by Alphabet's Q4 2025 earnings, and Gemini also powers AI Overviews reaching roughly 2 billion users a month inside Search. On raw model capability the two trade places often, but Google's advantage is that Gemini is embedded in Android, Chrome, Workspace, and Search rather than living in a standalone app.

Will AI answers kill Google's ad business?

So far, no. Alphabet's advertising revenue grew about 14% in Q4 2025 and Search revenue accelerated over the year. AI answers do compress the traditional click funnel, so Google is inserting ads directly into AI Overviews and AI Mode to keep monetizing. The long-term risk is real but has not shown up in the numbers yet.

What is the difference between AI Overviews and AI Mode?

AI Overviews are the Gemini-generated summaries that appear at the top of a normal results page above the usual links. AI Mode replaces the results page with a conversational, chatbot-style experience where you ask follow-up questions. Both now carry ads, and both are powered by Google's Gemini models.

How does Google DeepMind fit into all this?

Google DeepMind is the research group that builds the Gemini models after Google consolidated its AI teams under one roof. It ships the frontier systems, including the Gemini 2.5 family and reasoning features like Deep Think, that Google then distributes across its consumer products and sells to enterprises through Google Cloud.

Share this article: