Is Google's business in danger?

In brief: AI is changing clicks, where we ask questions and the cost of answering them. The author expects a transformation of Google's business rather than an immediate collapse, and proposes watching whether revenue growth leaves cash after investment.

Imagine you need a washing machine. You used to type a search into Google, open comparison sites and end up at a shop. Now you ask an assistant: “There are four of us, this is the space I have and I can spend 600 euros.” It suggests three models without opening a results page.

The need is the same. But between your question and the purchase, there used to be a journey that supported Google, comparison sites and shops. AI is shortening that journey. That is where this story begins.

Google charged for the click, and serving the search cost little

The machine worked like this: you searched for something. Google showed you results and ads. When you clicked an ad, the advertiser paid and Google collected. Since serving another search cost very little, it could keep a substantial share of that money.

That is why I care about the distinction between revenue and cash. Bringing in more money is not enough if you have to spend much more to earn it. Google's success came from combining two things: enormous demand and a cheap way to meet it.

It still works. In the third quarter, Google Search & other brought in $63,350 million, up 12% from a year earlier.[1] Paid clicks and their price are rising too.[2]

I do not see a business collapsing here. I see a thriving business being forced to change the mechanism that made it so profitable. AI is shifting all three parts at once: clicks, where we ask questions and the cost of answering them.

AI answers save steps and take away clicks

If the summary already answers my question, I have less reason to open a website. This happens within Google too: AI Overviews answers directly in search, and AI Mode lets me keep the conversation going without leaving it.

Pew observed clicks on traditional results in 8% of visits with an AI summary, compared with 15% without one. This is an observed association, not proof that the summary explains the entire difference.[3]

For whoever writes the washing machine comparison, the consequence is easy to understand: their information can help answer the question without the reader ever reaching their page. The answer becomes more useful; the website loses a chance to attract a visit.

A click through to a website and a click on an ad are not the same thing. Google says its summaries monetise at roughly the same rate as comparable searches.[4] It can keep me there and show me ads. But it has to make sure that this shorter journey still brings in money.

Some questions and their ads are already going elsewhere

The change does not stop within Google. I can open ChatGPT to choose the washing machine and return to the search engine for something else. I still use Google, but I have asked someone else the question that could lead to a purchase.

In September, ChatGPT reported 1.2 billion weekly users.[5] They have not all replaced the search engine. The figure shows the scale of another gateway to the internet.

And that gateway is already selling ads. OpenAI is testing them below answers, labelled and separate from them, according to its description.[6] In August, it reported an annualised advertising revenue run rate of around $1,000 million: this is a projection based on the pace of revenue, not annual revenue already earned.[7]

Google is also expanding ads in AI Mode and follow-up conversations.[8] The battle is still about appearing when someone decides what to buy. Only now, that decision can take shape in a conversation, far from the list of links Google dominated.

We do not have to stop using the search engine for this to matter. Some commercial questions only have to move elsewhere. The habit of opening Google can remain while part of the business leaves.

Defending that position means spending more than twice as much

Google has to offer AI answers to keep those questions. But generating them takes much more computing power than serving a conventional search. Even if it manages to earn the same from advertising, it cannot assume it will have the same amount left after paying for the infrastructure.

Alphabet expects to invest around $200,000 million this year, compared with around $91,400 million last year: more than twice as much.[9][10] The investment covers the whole group, not just Search, but it shows the scale of the effort needed to sustain this new phase.

The money goes out now; the accounting cost of the equipment is spread over years. That is why revenue can grow and margins can hold up while the cash available after investment shrinks.[1][2]

If I subtract capital expenditure from cash generated by operations, quarterly free cash flow comes out at negative $9,065 million, compared with positive $24,461 million a year earlier. This is a calculation based on Alphabet's accounts.[1]

Payments for share buybacks were zero in the quarter.[1] That figure alone does not explain the decision, but it reinforces the contrast with the old abundance of cash.

Google says it is reducing the cost of its AI answers.[11] It may become more efficient. Even so, it does not publish a cost per answer or a margin specifically for that kind of search. I want to see whether the new machine ends up leaving as much cash as the one it is replacing.

That is the bet: spend to keep the user, then work out how to make the answer profitable. I understand why Google is making it. Leaving questions to a rival would also put its advertising at risk.

Reddit shows what happens when visits stop coming

While Google tries to hold on to the business, those who depended on its traffic face the change from a weaker position. Reddit lets us see this without inventing a collapse: its quarterly revenue grew 53%.[12]

At the same time, it said that in October it was receiving, on average, more than one million fewer weekly visits from search engines. These are traffic referrals, not distinct people.[13] If part of the conversation is settled before anyone enters Reddit, growth today does not guarantee that the audience will keep coming tomorrow.

Following the traffic warning and the company's outlook, its shares fell around 29%, according to Reuters.[14] The reaction reflects that combination of news; it does not allow us to attribute the entire fall to AI.

The lesson I take from this is simple: rising revenue and an underlying problem can coexist. For websites that depended on visits, the danger is losing them. For Google, it is keeping the question but earning less from it after paying for the answer.

The law is already watching who controls the new gateway

Judge Mehta in the United States, the UK's CMA and the European Commission under the DMA are already looking at AI as the next gateway: who controls the answers and access to content and data matters.[15][16][17] The news item from 31 October addresses another piece of the puzzle: preferences about training.[18] Expressing them does not guarantee visits or payment.

Google can keep growing and lose its old cash machine

My forecast for the next two years is an uneasy coexistence. Google will remain an important gateway, assistants will capture more decisions and infrastructure will absorb a lot of cash. I expect a transformation of the business, not an immediate collapse.

Is Google in danger today? I would say no. Is the machine that let it earn so much from each search under threat? Yes. To see whether it can defend it, I will watch whether revenue grows without investment consuming the cash, and whether it manages to earn money from the decisions now taking place through assistants.

Back to the washing machine. I want to choose well and buy it with less effort. If an assistant can do that for me, I will use it. Google needs to keep earning money along that journey and have money left over after serving me. That is the calculation that will decide its future, even if it is selling more ads today.