ChatGPT Still Leads, but the Moat Is Thinner Than It Looks

## ChatGPT Still Leads, but the Moat Is Thinner Than It Looks  On May 14, 2026, [Decrypt](https://decrypt.co/367884/chatgpt-losing-ground-anthrophic-google-numbers) published a piece about ChatGPT losing ground to rivals. The cleanest number behind that story comes from the January Global AI Tracker summarized by [Search Engine Journal](https://www.searchenginejournal.com/google-gemini-gains-share-as-chatgpt-declines-in-similarweb-data/564690/): ChatGPT accounted for 64.5% of gen-AI chatbot site visits, while Gemini held 21.5%. A year earlier, the same tracker had ChatGPT at 86% and Gemini at 5%. That is not a collapse. It is a market that is starting to look less monogamous. The reason that matters is simple. A category leader can still be the category leader and still be losing incremental mindshare at the edges. In web traffic terms, the center is not disappearing. It is just no longer the only place where users are likely to start. ### 64.5% versus 86% is the change that matters When a leader drops from 86% to 64.5% in a year, the lazy reaction is to treat it as a product failure. That misses the more useful reading. The category has become crowded enough for distribution to matter as much as quality. Some users are following Android and Workspace surfaces into Gemini. Others are meeting Anthropic in enterprise workflows. A smaller but noisy group is moving through X, where Grok benefits from attention loops and bundled placement that have little to do with benchmark charts. The same January tracker also showed that total category traffic softened over the winter break. That matters because it separates two things that are easy to blur: a seasonal dip in total visits and a real change in share. ChatGPT can still be the biggest player and still be losing incremental mindshare at the edges. ## Why this looks more like distribution change than category collapse The strongest reason to stay cautious is methodological. Similarweb's tracker measures direct domain visits. It does not count API traffic, embedded assistants, or the many places where AI is already hidden inside products people open for other reasons. So the tracker is useful, but it is a view of the doorway, not the whole house. That limitation cuts both ways. It means ChatGPT's lead may be larger than the web number suggests. It also means Gemini's rise may be more durable than a simple traffic chart implies, because Google can route users through Android, Chrome, Gmail, and Workspace rather than asking them to remember a separate destination. Fortune's February coverage pointed in the same direction: Gemini.google.com had 28.38% more traffic than ChatGPT.com in December 2025, while ChatGPT traffic fell 5.59%. The same report said Grok's share had climbed to 15.2% from 1.6% a year earlier. That is the kind of detail that tells you this is not a single-rival story. It is a distribution war. ### What the tracker cannot see There are three blind spots that matter. - API usage can grow even when direct web visits slow down. - Enterprise adoption can move through pilots, contracts, and embedded workflows before it ever shows up as a clean consumer-web trend. - Bundled assistants can gain share without looking like a separate app win, because the user never experiences the switch as a search for an AI tool. That is why the market should be careful with victory laps and panic headlines. The right question is not whether ChatGPT is "done." The right question is whether its usage is becoming less central to how people access AI. ## Gemini is the cleaner pressure point, Grok is the noisy one Gemini's rise is the cleaner signal because it sits inside Google's product stack. If Gemini keeps converting search, Android, and Workspace exposure into repeat usage, the share shift has a structural path. ChatGPT can still dominate on brand and habits and still lose direct visits where default placement matters. Grok deserves attention, but for a different reason. It is less a clean rival story than a distribution experiment wrapped around X. A rise in share there may say as much about attention loops and platform bundling as it says about product preference. That does not make it unimportant. It makes it harder to generalize from. The more useful read is that ChatGPT is no longer facing a single challenger. It is facing several different kinds of pressure at once: Google's default distribution, Anthropic's enterprise positioning, and smaller products that can spike when the audience is already inside a parent platform. ## What would prove the shift is durable There are three signals worth watching. - Direct web share keeps drifting away from ChatGPT after seasonal distortions fade. - Gemini converts distribution into repeat usage instead of one-time trial traffic. - Anthropic continues to take enterprise mindshare without needing a consumer app breakthrough first. If those three move together, the market is not just rearranging rankings. It is changing the way users enter AI products in the first place. If they do not, then this is still a leader with a narrower edge, not a leader in retreat. The safest conclusion is also the least dramatic one: ChatGPT remains the biggest name in the category, but the category is now mature enough that default distribution, enterprise workflow, and bundled surfaces matter as much as model quality. --- Author: [Alex Chen](https://x.com/AlexC0in) | Alex has followed blockchain technology since 2021, focusing on DeFi and on-chain data analysis Source: [decrypt.co](https://decrypt.co/367884/chatgpt-losing-ground-anthrophic-google-numbers)