ChatGPT Falls Below 50% Market Share for the First Time as Gemini and Claude Surge

For nearly four years, ChatGPT has dominated the AI assistant market so comprehensively that competitors measured their success against it in percentage points. That era has not ended, but its most defining data point just shifted: according to Sensor Tower’s newly released State of AI 2026 report, ChatGPT’s share of global AI assistant usage fell below 50% for the first time in March 2026, reaching 46.4% by May. The milestone is less an inflection point than the confirmation of a structural trend that has been gathering momentum for over a year.

The Numbers

Sensor Tower measures what it calls “true audience” — a cross-platform metric combining desktop, mobile apps, and mobile web usage — giving it broader coverage than figures based solely on app downloads or web visits.

The snapshot at May 2026:

• ChatGPT: 46.4% market share, 1.1 billion monthly active users
• Google Gemini: 27.7%, approximately 662 million MAU
• Claude: 10.3%, approximately 245 million MAU

Together, ChatGPT, Gemini, and DeepSeek control roughly 90% of time spent on AI assistant applications globally in Q1 2026. The remaining 10% is distributed across a long tail of smaller assistants, including Grok, Perplexity, Meta AI, and dozens of regional platforms.

The velocity of change is arguably more striking than the current snapshot. Claude’s true audience grew 452% year-over-year through May 2026, with US market share rising from 4.4% to approximately 14% in the same window. Gemini’s monthly engagement per user surged from 14 minutes to 100 minutes over the past year. ChatGPT continues to grow in absolute terms — crossing 1.1 billion monthly active users, the fastest mobile app ever to reach that figure — but at a pace that competitors are now matching or exceeding.

What Is Driving the Shift

Three forces are reshaping the competitive landscape in ways that favor incumbents with massive platform leverage.

Android integration. Gemini’s growth is substantially tied to its embedding in Android 17, where it functions as OS-level infrastructure rather than a standalone app. With over 3 billion Android devices active globally, Google can reach users at the system level without requiring them to make a conscious choice to switch assistants. Gemini’s engagement per user growing from 14 to 100 minutes monthly is almost certainly connected to this ambient availability — users encounter Gemini in their keyboard, camera, notifications, and search before they ever open the Gemini app itself.

Subscription conversion. Claude stands out in the Sensor Tower data not for raw user numbers, but for paid conversion. Claude’s 13% conversion rate for paid subscriptions is the highest in the industry, significantly above ChatGPT’s rate. This means that while Claude has far fewer users than ChatGPT or Gemini, a disproportionately high fraction of its users are paying customers — suggesting strong product-market fit among the professional and developer segments that Anthropic has targeted most aggressively. Claude Code, Anthropic’s AI coding assistant, appears to be a significant driver of both engagement and paid conversion.

OpenAI’s strategic complications. The Sensor Tower data arrives against a backdrop of escalating legal and political pressure on OpenAI. A 42-state attorney general investigation opened in June, and the company’s Defense Department contracts triggered a measurable uninstall spike documented in the same report. OpenAI’s advertising spend increased 800% year-over-year as it works harder to maintain its user base — a sign that organic growth is no longer sufficient to sustain its market share position.

The Engagement Gap Closes

Perhaps the most structurally significant finding in the Sensor Tower report is the shift in engagement depth, not just user counts.

ChatGPT users average approximately 215 minutes of usage per month. Claude users now average 120 minutes, up from 40 minutes a year ago. Gemini users average 100 minutes, up from 14. The gap between the leader and its competitors on this metric has compressed dramatically — and engagement per user is arguably a better predictor of long-term revenue potential than raw monthly active user counts.

The e-commerce section of the Sensor Tower report adds another dimension. GenAI referral traffic to major retail sites has increased sharply, with the electronics category seeing traffic from AI assistants nearly quadruple. Amazon’s Rufus AI shopping assistant drives purchase conversion at nearly twice the rate of non-AI assisted browsing. This commercial integration — AI assistants not just answering questions but completing transactions — represents the next phase of competition, and it is one where Google, with its Shopping and Maps infrastructure, and Amazon, with its marketplace, have structural advantages that neither OpenAI nor Anthropic currently match.

The Ad Spend Signal

One of the more underreported angles in the Sensor Tower data is the advertising picture. AI-related ad spending reached $1.3 billion in Q1 2026 alone, with 167 billion impressions. OpenAI’s advertising increased 800% year-over-year; Anthropic’s rose 1,184%. The global GenAI ad spend more than tripled in Q1 2026 versus the prior year.

This level of marketing investment is consistent with a market where the equilibrium has not yet settled — where all major players believe they can gain or defend share through acquisition spending. It also suggests that the top-of-funnel differentiation between assistants may be narrowing: when products become increasingly similar in capability, advertising becomes a more important lever.

What Comes Next

The structural trajectory points toward continued share loss for ChatGPT at the top-line level, counterbalanced by its formidable absolute user base and the premium monetization potential of its ChatGPT Business offering. The company’s IPO, expected in September 2026, will put the market’s assessment of these dynamics in stark financial relief for the first time.

For Google, the story is about converting Android’s structural reach into genuinely engaged Gemini users — and the engagement data suggests that conversion is happening, albeit from a low base.

For Anthropic, the story is fundamentally different: it is competing not for the largest user base but for the most commercially valuable users. A 13% paid conversion rate on 245 million users is a different business model than a 5% conversion rate on 1.1 billion. Which approach produces better long-term unit economics will depend on how AI assistant pricing evolves over the next two to three years.

What the Sensor Tower report confirms is that the AI assistant market in 2026 is genuinely competitive in a way it was not in 2024. The winner-takes-all dynamic that once seemed inevitable has given way to something considerably more interesting.