AI Is Much Bigger Than You Think

When most people try to measure how big AI has become, they compare website traffic, typically ChatGPT.com visits, against Google.com visits. It is an instinct. It is also wrong, and the error is not small.

New analysis from Graphite.io puts a number on that error: web-only comparisons underestimate AI usage by 4 to 5 times. Once you account for mobile apps, where the vast majority of AI interaction actually happens, the picture looks dramatically different.

Key Numbers at a Glance

45 billion  Monthly AI sessions worldwide across major LLM platforms  (global total, 2026)

56%  Of global search engine volume, that is what AI now equals  (Q4 2025)

83%  Of all global AI usage happens inside mobile apps, not browsers  (worldwide)

~300%  Year-over-year growth in U.S. AI usage by December 2025  (U.S. only)

These numbers come from a methodology that combines web traffic and mobile app session data across the five largest LLM products, ChatGPT, Gemini, Perplexity, Grok, and Claude, and compares them with the six largest search engines. That is a meaningfully different calculation than pulling data on two websites.

Why the Standard Comparison Fails

The conventional benchmark, ChatGPT web traffic versus Google web traffic, captures a small slice of actual AI usage. It misses mobile apps almost entirely. And for AI platforms, mobile is where the majority of interaction occurs.

Where AI sessions actually happen

This matters because when analysts compare ChatGPT.com to Google.com, they are looking at the browser layer of AI usage only. Subtract that and you have excluded three-quarters of American AI interactions and more than four-fifths of global ones.

The typical web-only comparison is not a close approximation, it is a systematic undercount. Correcting for it changes the conclusion from "AI is an emerging channel" to "AI is already a major channel."

What 45 Billion Sessions Actually Means

The five major LLM platforms combined generate approximately 45 billion monthly sessions globally. Set against roughly 80 billion monthly sessions across the six largest search engines, that puts AI at 56% of global search volume, more than half.

In the United States, the figure is lower, around 34%, but growing faster. U.S. AI usage was up approximately 300% year-over-year by December 2025. Global usage has plateaued somewhat since mid-2025 as the first wave of adoption normalises, but U.S. growth remains steep.

AI vs. Search: Estimated Global Monthly Sessions

A key nuance: not all AI prompts are equivalent to search queries. Research from OpenAI suggests around 52% of prompts are information-seeking, the closest analogue to a traditional search query. The remainder are task-oriented or expressive. If you isolate only those information-seeking interactions, AI accounts for roughly 28% of global search volume and 17% in the U.S., still a significant portion, and one that is growing.

ChatGPT Dominates, But the Landscape is Fragmented

Within the AI channel itself, usage is not evenly distributed. ChatGPT accounts for the majority of global AI sessions, but the total activity across all platforms is what creates the strategic picture.

AI platform session share (global estimate)

The dominance of ChatGPT is a useful starting point for prioritisation, but it does not mean the other platforms can be ignored. A brand researched on Perplexity reaches a different audience, typically more research-oriented and higher intent, than one found on ChatGPT. What each platform says about your brand can differ substantially, and what it says matters to the buyer who sees it.

Search Is Not Shrinking, The Total Pie Is Growing

One of the more counterintuitive findings in the data is that AI's growth has not come at the direct expense of search. Total usage across both channels has grown approximately 26% globally since 2023. AI is expanding the discovery landscape, not just redirecting existing queries.

This is not a question of abandoning SEO for GEO. It is a question of adding AI visibility as a parallel priority. Brands that treat them as mutually exclusive are misreading the market.

Your buyers are now conducting discovery across two distinct channels that operate by different rules, draw on different data, and require different strategies. Winning in one does not automatically mean winning in the other.

Google's Share Is Declining, Slowly But Measurably

Even though total search volume is growing, Google's share of search-related activity is falling. In 2023, Google accounted for roughly 89% of all search-related sessions globally. By Q4 2025, that figure had dropped to around 71%. That shift represents a meaningful redistribution, one that is likely to continue.

Google's share of search-related activity (global)

For brands that have built their entire discovery strategy around Google rankings, this should prompt a clear question: what happens to your visibility as that 18-point share shift continues? The brands best positioned for the next few years are the ones investing in AI visibility now, before the redistribution accelerates further.

What This Means for Your Brand

The data points to four practical implications that should shape how brand and marketing teams think about AI visibility in 2026.

01. Your buyers are already there

45 billion monthly sessions is not a future scenario, it is the current state. Your buyers are using AI platforms to research products and services right now. If your brand is not showing up accurately in those answers, a competitor is.

02. Mobile is the primary surface

83% of AI sessions happen in mobile apps. This means the standard web traffic comparison that most teams use to assess AI impact understates it by 4 to 5 times. The channel is far more active than most dashboards suggest.

03. SEO does not equal AI visibility

Ranking first on Google for a query does not mean you appear in AI answers for the same query. The two channels use different signals, draw on different sources, and require separate strategies. Conflating them is a planning error.

04. The channel is fragmented

ChatGPT, Gemini, Perplexity, and others each construct answers differently. Visibility on one does not transfer to another. Each platform needs to be audited and tracked independently, with consistent measurement, not one-off checks.

Measuring What Actually Matters

Most teams that are aware of AI visibility are not yet measuring it systematically. They might run occasional manual checks, asking ChatGPT about their brand and reviewing what comes back, but that approach captures a snapshot of a single model on a single day. It does not track how representation changes over time, how it varies across models, or how it compares to competitors.

Effective AI visibility measurement requires tracking brand mentions across platforms, monitoring competitive share of voice, understanding which sources AI models cite when referencing your brand, and identifying where inaccuracies or gaps exist.

The scale of AI usage is no longer a question. The question is whether your brand's presence in that channel is managed or left to chance.