ChatGPT Just Slipped Below Half the Assistant Market

For two and a half years, "the AI assistant market" and "ChatGPT" were close enough to the same thing that the distinction didn't matter. That era just ended quietly. According to Sensor Tower's State of AI 2026 report, ChatGPT's share of global AI-assistant usage fell below 50% for the first time, slipping to 46.4% by May. It is still, by a wide margin, the single biggest assistant on earth—the report also clocks it hitting one billion monthly active users in record time. But "biggest" and "majority" are no longer the same word, and the gap between them is where the next two years of this market will be fought.

The headline number is easy to misread as decline. It isn't. ChatGPT's audience is still growing in absolute terms; what's changed is that the rest of the field finally started growing faster. The crossover under 50% is less a story about OpenAI losing users than about the market maturing into something with more than one default.

Who's taking the other half

Sensor Tower's breakdown for May 2026 puts Google Gemini at 27.7% and Anthropic's Claude at 10.3%. Those two numbers tell very different stories about how share gets won.

Gemini's climb is the distribution story. Its gains are concentrated where Google already owns the surface area—Android integration, the default assistant slot, AI Mode bleeding into Search, and broad ecosystem placement across Google's apps in Europe, the US, Japan, and South Korea. When the assistant is one swipe from the home screen on billions of phones, you don't need to win an argument about model quality; you need to be there, and Google increasingly is. A meaningful share of Gemini's 27.7% is users who never made an active choice at all, which is exactly the kind of share that's hard to dislodge.

Claude's climb is the opposite—and arguably the more remarkable line in the whole report. Sensor Tower puts Claude's "True Audience" up 452% year-over-year in May, with US share rising from 4.4% to nearly 14% over the same window. That is not distribution; Anthropic has no phone, no browser, no operating system to ride. That growth is pull, not push: users seeking it out, mostly for coding and serious knowledge work, and bringing it into workflows it wasn't pre-installed in. A 452% jump off a small base is easy to discount, but the direction and the reason—earned preference rather than placement—is the part competitors should find unsettling.

The 90% that didn't move

Here's the counterweight to all the share-shuffling: Sensor Tower also reports that ChatGPT, Gemini, and DeepSeek together account for roughly 90% of total time spent across AI-assistant apps in Q1 2026. The market is concentrating and fragmenting at the same time—the top is splitting three (or four) ways, but the long tail of also-ran assistants is being crushed flat underneath it.

That's the real competitive picture. The fight isn't "ChatGPT vs. everyone." It's a three-to-four-horse race—OpenAI, Google, Anthropic, and DeepSeek as the open-weight wildcard—inside a market where everyone else is rounding error. Falling below 50% doesn't move ChatGPT toward the pack; it just means the pack now has named members instead of one leader and a blur.

Why share is the wrong thing to watch

The more interesting thread buried in the report is that the operators may have stopped optimizing for the share number entirely. Sensor Tower frames 2026 as the year the industry pivots from land-grab to monetization—and specifically toward AI quietly rewiring commerce, with assistants becoming a discovery and transaction layer rather than just a chat box.

That reframes everything. If the prize is becoming the place where people decide what to buy, then raw assistant share matters less than which assistant sits closest to a purchase. Gemini's Search and shopping adjacency looks very different through that lens. So does OpenAI's push into agentic commerce. The market-share table measures attention; the monetization shift measures intent, and intent is where the money actually is.

What it means

For builders, the practical takeaway is that "just ship on ChatGPT" stopped being a complete distribution strategy somewhere this spring. A platform at 46% means a real—and growing—slice of your users live in Gemini or Claude by default, and those platforms reward different things. Gemini rewards being present in Google's ecosystem; Claude rewards being genuinely good at the hard task a power user came for. Betting on one surface is now betting against half the market.

For the labs, the under-50% line is a useful tombstone for an assumption that quietly ran the industry since 2023: that incumbency in this market was self-reinforcing and roughly permanent. It wasn't. Claude's 452% says preference can still be earned at speed; Gemini's 27.7% says distribution can still be leveraged at scale. ChatGPT is still the giant. It's just no longer the whole game—and the report makes clear it never really feared losing users so much as it should fear losing the next layer, the one where assistants stop being something you talk to and start being the thing you buy through.