Google just posted its best quarter ever. OpenAI just launched a browser to replace it. Both things can be true, and the implications for advertising, commerce, and payments are enormous.

On October 21, 2025, OpenAI launched Atlas, a Chromium-based web browser where ChatGPT's search bar greets you on open, not Google's. The pitch was not subtle. Atlas puts ChatGPT at the centre of the browsing experience from the moment you launch it: search, shopping, email, research, all routed through AI first. Fortune described it as a bid to supplant Google as the internet's universal starting point.

Three months later, Alphabet dropped its Q4 2025 earnings. Search revenue grew 17 percent year over year to $63.1 billion in a single quarter. Total revenue hit $113.8 billion, up 18 percent. Earnings per share came in at $2.82, well above analyst expectations. It was, by most measures, the strongest quarter in the company's history.

Two opposite signals. Same industry. Same quarter.

If you read our analysis of the $650 Billion Squeeze, you know the AI infrastructure arms race is accelerating. Google's Q4 results are the first real data point showing whether that spending is generating returns. The answer is more complicated than either side wants to admit.

Start with the numbers that made the headlines. Google's Q4 was exceptional across the board. Google Cloud revenue grew 48 percent to $17.7 billion, with operating margins expanding from 17.5 percent to 30.1 percent. Cloud backlog reached $240 billion, up 55 percent sequentially. Gemini now has over 750 million monthly active users, with 8 million paid Enterprise seats sold in just four months. And Alphabet reduced Gemini serving costs by 78 percent over the course of 2025.

Those numbers tell a story of momentum. But there is another set of numbers that tells a different story.

In February 2024, Gartner predicted that traditional search engine volume would drop 25 percent by 2026, with AI chatbots and virtual agents absorbing the queries. Apple's SVP of services testified in court that search queries on its devices declined for the first time ever. Perplexity has crossed $200 million in annualised revenue and handles roughly 780 million queries per month. And according to Press Gazette, Google search traffic to publishers declined globally by a third in the year to November 2025, with US traffic specifically down 38 percent.

Google's revenue is growing. But the ecosystem that generates that revenue is eroding underneath it. The question is not whether the cracks exist. It is how long the surface holds.

The resolution to this paradox is that search is no longer one market. It is two.

The first market is informational. Research queries, how-to questions, fact-checking, definitions, and the kind of open-ended exploration that used to fill the first three pages of Google results. This is the market that AI is winning decisively.

And it makes sense. If you can ask a question and get a direct, synthesised answer without clicking through ten links, reading three articles, and mentally triangulating the truth, why wouldn't you?

The second market is transactional. Product searches, price comparisons, local business lookups, travel bookings, and anything that ends in a purchase rather than a paragraph. This is the market where Google still dominates. Transactional queries have commercial intent attached to them, and Google has spent two decades building the infrastructure to capture that intent and convert it into revenue.

The numbers confirm the split. According to First Page Sage, ChatGPT holds roughly 64 percent market share for creative and intellectual queries, while Google retains over 90 percent for transactional searches and 93 percent for navigational queries. They are not competing for the same users in the same moments anymore.

This is why Google's revenue can grow 17 percent while the broader search ecosystem deteriorates. The informational queries were never the ones paying the bills. But here is the problem: they were the ones building the habits. Every time a user went to Google to ask a question, they were one step closer to going to Google to buy something. The top of the funnel fed the bottom.

And the competitors are not stopping at answers. Perplexity launched free agentic shopping with PayPal integration in November 2025, letting users complete purchases from over 5,000 merchants without leaving the conversation. Opera shipped Neon, a browser where an agent called "Neon Do" can shop, book, and transact directly inside your live browser session, no cloud handoffs, no shared passwords. These are not research tools encroaching on search. They are commerce tools that happen to start with a question.

If the informational layer migrates permanently to ChatGPT, Perplexity, and Atlas, the transactional layer does not collapse overnight. But the pipeline that sustains it starts to thin. And the competitors building on top of that informational layer are already reaching for the transaction.

This is where the advertising model faces its most serious structural challenge.

Google's search advertising has always been a shotgun. Broad keywords, ten blue links, and a bet that somewhere in the scatter of results, a user clicks on something that generates revenue. It worked brilliantly when Google was the only place people went to ask questions. Every query, informational or transactional, passed through the same funnel. Advertisers paid for placement. Publishers provided the content. Users provided the attention. Everyone got their cut.

AI is a rifle. Specific intent, precise answers, no wasted motion. And the data on what happens when that rifle shows up is stark.

According to Seer Interactive, organic click-through rates on queries featuring AI Overviews fell 61 percent between mid-2024 and September 2025, dropping from 1.76 percent to 0.61 percent. Paid click-through rates on those same queries plunged 68 percent, from 19.7 percent to 6.34 percent. And the decline is not limited to queries with AI Overviews. Even queries without them saw organic CTR fall 41 percent year over year.

The zero-click trend tells the same story from a different angle. According to Similarweb data, zero-click searches increased from 56 percent to 69 percent between May 2024 and May 2025. On queries where AI Overviews appear, the zero-click rate hits 83 percent. Eight out of ten users get their answer without ever leaving the search page.

Publishers are already feeling this. Press Gazette reported that Business Insider saw organic search traffic fall 55 percent between April 2022 and April 2025. HuffPost lost half its search referrals over the same period. HubSpot reported declines of 70 to 80 percent. Media executives surveyed expect traffic to decline by 43 percent on average over the next three years, with a fifth expecting losses exceeding 75 percent.

And here is what should worry Google most: Gemini has not offset this in any meaningful way. If anything, AI Overviews are accelerating the erosion by training users to consume answers inside the search page rather than clicking through to the sites that create the content. Google is cannibalising its own advertising ecosystem. The content that fills AI Overviews comes from publishers. The clicks that pay those publishers are disappearing because of AI Overviews. The feedback loop is destructive, and there is no clear mechanism to reverse it.

The creative, conversational AI services will only grow from here. Every user who discovers they can get a direct, specific answer from ChatGPT or Perplexity instead of wading through a page of links and ads is a user who may not come back for the informational query. And over time, that habit shift bleeds into the transactional queries too.

Google is not standing still. If the link-click-ad model is under pressure, the company is building an alternative: own the transaction itself.

In January 2026, Google and Shopify jointly announced the Universal Commerce Protocol at the National Retail Federation conference. UCP is an open-source standard that allows AI agents to connect directly with merchant backends to browse inventory, negotiate terms, and complete purchases. The list of backers reads like a payments industry roll call: Visa, Mastercard, Stripe, Adyen, American Express, PayPal. Retailers include Target, Walmart, Best Buy, Macy's, The Home Depot, and Zalando.

The integration goes deep. According to the Google Developers Blog, UCP will power native checkout inside AI Mode on Google Search and the Gemini app, allowing users to complete purchases without ever leaving Google. Google Pay handles the credential layer, with PayPal integration coming. Shopify has built Agentic Storefronts into its admin panel, letting merchants manage AI-powered sales across Google, ChatGPT, and other surfaces from one dashboard.

This is the pivot from search engine to commerce infrastructure. Instead of earning a fee when someone clicks an ad, Google earns a position in the transaction chain when someone buys something. The shift from "search and click" to "search and buy" without leaving the platform.

OpenAI is making the same play. Shopify's Agentic Storefronts connect to ChatGPT and Atlas too. Stripe is embedded in both ecosystems. Perplexity's agentic shopping already handles checkout through Firmly.ai and PayPal, with merchants joining through a single API. Opera Neon's "Do" agent executes purchases directly in the browser session where the user is already authenticated.

And OpenAI is now reaching for the advertising revenue too. This week, ChatGPT began testing ads in its answer engine, with Target and its retail media division Roundel among the first partners. The competition is no longer limited to who answers the question or who completes the purchase. It now includes who gets paid for the attention in between.

For the payments industry, this is where the story gets tangible. Every checkout inside an AI surface still needs a payment rail. That is why Visa, Mastercard, Stripe, and Adyen are all backing UCP. UCP's security architecture tokenises payment and identity information through credential providers, while payment service providers handle transaction processing separately. Agents never touch raw payment data. The payments infrastructure does not care whether the purchase was initiated by a human clicking a button or an AI agent negotiating a deal. It just needs to be in the path.

The question is not whether there will be a payment at the end. It is who controls the path to it.

This brings us back to the $650 Billion Squeeze.

Alphabet announced 2026 capital expenditures in the range of $175 billion to $185 billion. At the top end, that is more than double the $91.4 billion it invested in 2025 and exceeds its total spending for the previous three years combined. Analysts had expected $119.5 billion. Meta announced its own range of $115 billion to $135 billion. The scale of investment is unprecedented.

And the market's reaction was immediate. Alphabet shares dropped 7 percent the day after earnings despite beating on revenue and earnings. Deutsche Bank analysts said Alphabet had "stunned the world" with its capex plan. CEO Sundar Pichai told analysts that compute capacity constraints keep him up at night and that the company expects to remain supply-constrained throughout 2026.

The spending is not slowing down. On February 10, Alphabet raised nearly $32 billion in debt in less than 24 hours across sterling, Swiss franc, and dollar bond sales, including a 100-year note, the first by a technology company since the dot-com era. And in its latest annual filing, Alphabet disclosed new AI-related risks, explicitly acknowledging that consumers adopting generative AI may "decrease their use of internet search" and that the company could end up with "excess capacity" from its infrastructure commitments.

The bet is straightforward: AI does not replace search, it transforms search from an information tool into a commerce platform. If that thesis proves out, the $185 billion is a down payment on owning the transaction layer of the internet. Cloud revenue grows. Commerce fees grow. The advertising decline becomes a manageable transition, not a crisis.

If it does not prove out, the numbers get uncomfortable quickly. Search advertising still represents roughly 56 percent of Alphabet's total revenue. If that revenue dips even modestly for two consecutive quarters, while $185 billion in capital commitments are rolling out, the stock reaction will not be a 7 percent dip. It will be a correction. And corrections trigger knee-jerk decisions: cost cuts, strategy pivots, hiring freezes, and the kind of short-term thinking that undermines long-term bets.

The software sector has already lost 30 percent of its value in the last three months on AI spending concerns. Investors are not patient. The window between "visionary investment" and "reckless spending" is measured in quarters, not years.

If the advertising model that funded the open web for two decades is being hollowed out by the very AI it helped build, who pays for the internet that comes next?