ChatGPT Just Lost Its Majority. The Real Story Is Ads, Not Decline.

TL;DR

• ChatGPT's share of global AI-assistant users fell to 46.4% by the end of May 2026 , its first time below 50% , per Sensor Tower's State of AI 2026 report (released June 16).
• It is still #1 by raw scale: 1.1 billion+ monthly users vs Gemini's 662M (27.7%) and Claude's 245M (10.3%), per Sensor Tower.
• This is not decline , it's the market outgrowing one app. H1 2026 saw 2.3 billion AI-app downloads and $4.2B in spend (up from $1.83B a year earlier).
• The under-covered story: OpenAI is now serving ads to ~17% of daily ChatGPT users (up from a February test), with Walmart, Target, and Costco shopping integrations.

ChatGPT just lost its majority for the first time since it created the category in 2022. But "ChatGPT falls below 50%" is the wrong headline to fixate on , the app added users faster than ever and still doubles its nearest rival. The two things that actually matter for anyone building on or marketing through AI are quieter: the assistant layer is fragmenting into a real multi-model market, and OpenAI has started turning that attention into an ad business.

What the numbers actually say

Per Sensor Tower, ChatGPT's "true audience" share (measured across desktop, mobile apps, and mobile web) went 65.3% in December 2024 → 52.8% in December 2025 → 46.4% by May 2026, crossing below 50% in March. Yet in absolute terms it crossed 1.1 billion monthly active users in May , the fastest app in history to that mark, ahead of TikTok, YouTube, and Instagram. Both things are true: ChatGPT is bigger than ever and a minority of a much larger market.

The share moved because the field grew. Gemini sits second at 27.7% (662M users), Claude third at 10.3% (245M), and the rest , Meta AI, Grok, Perplexity, DeepSeek , collectively ~15.6%. The standout is Claude: Sensor Tower clocked its true audience up 452% year-over-year in May, with US share climbing from 4.4% to nearly 14%.

The real story #1: the assistant layer is now multi-model

For two years "AI strategy" meant "ChatGPT strategy." That's over. A market where the leader holds 46% and two credible challengers hold a combined ~38% is a market you can no longer address through one model. If you're optimizing content for AI visibility, building on an assistant API, or thinking about where your customers ask questions, the answer is now plural , Gemini and Claude are not rounding errors. Single-model dependence is becoming a real risk, both for cost and for reach.

The real story #2: ChatGPT is becoming an ad network

OpenAI started testing ads in ChatGPT in February and has scaled steadily , by May, an average of 17% of daily users were seeing them, per Sensor Tower. The largest advertiser categories so far are software and shopping, then media/entertainment and food/dining, and OpenAI has wired in shopping with Walmart, Target, and Costco. That is the beginning of a new, enormous ad surface , one that sits inside the answer, not beside it. For marketers, "being recommended by ChatGPT" is shifting from a pure-organic game toward a paid one. For users, the ad-free era that drove adoption is ending, which itself adds switching friction (a tailwind for Gemini and Claude).

The trust variable nobody prices in

One data point worth holding onto: when OpenAI announced a US Department of Defense deal in February 2026, Sensor Tower recorded a measurable spike in ChatGPT uninstalls and a matching surge in Claude downloads. In a multi-model market with low switching cost, brand values move share. That's new , and it favors whichever provider a given audience trusts most.

What this means for you

• Stop optimizing for one model. Treat AI visibility as multi-model , test how Gemini and Claude answer your category, not just ChatGPT.
• Plan for paid AI placement. Ads inside ChatGPT answers are coming at scale; budget and positioning will follow, the way they did for search.
• If you build products: avoid single-provider lock-in. Pricing, availability, and trust now vary enough across ChatGPT/Gemini/Claude that portability is leverage.
• Watch the OpenAI S-1. OpenAI filed confidentially on June 8, 2026; the ad ramp is partly a pre-IPO revenue story, so expect it to accelerate.

Frequently asked questions

Did ChatGPT actually lose users?

No. ChatGPT grew to 1.1 billion+ monthly users by May 2026 , a record. Its share fell to 46.4% because the overall AI-assistant market grew faster, led by Gemini and Claude.

Who is gaining the most?

Claude. Sensor Tower reported its true audience up 452% year-over-year in May, with US market share rising from 4.4% to nearly 14%. Gemini remains the #2 overall at 27.7%.

Is ChatGPT really showing ads now?

Yes. OpenAI began testing ads in February 2026 and, per Sensor Tower, ~17% of daily users were being served ads by May, with shopping integrations including Walmart, Target, and Costco.

What should marketers do about it?

Go multi-model (test Gemini and Claude, not just ChatGPT) and prepare for paid placement inside AI answers as ChatGPT's ad surface scales.

Sources

• TechCrunch , ChatGPT's market share slips below 50% for the first time
• Sensor Tower State of AI 2026 report (summary)
• Fast Company , ChatGPT loses ground to Gemini and Claude

Related reading

• Real-time LLM Inference on Standard GPUs: 3k tokens/s per request
• Claude Fable 5 Scores 95% on Its Own Benchmark and 19% on Real Security Work. The Gap Is the Lesson.
• I Tracked 592 Signals Across 6 B2B Niches for 3 Weeks. AI Won — But Not How You'd Expect