AI Chatbot Race:

Can Anyone Catch ChatGPT? Study Finds 2037 Might Be the Year

We analyzed monthly web visits for ChatGPT, Claude, Gemini, Grok, and DeepSeek from August 2025 through August 2026. The data was sourced from OneLittleWeb’s Most Visited Websites data dashboard, with web visit estimates gathered by OneLittleWeb.com from Ahrefs and Semrush.

Source: https://onelittleweb.com/digital-market-intelligence/popular-websites/

Web visits moved up and down sharply from month to month. So instead of applying one fixed growth rate, we modeled the monthly changes themselves.

We used a first-order autoregressive model, or AR(1). This lets one month’s change in web visits influence the next month’s change. We then resampled the historical “shocks”, the part of each month’s change the model could not explain, and ran 250,000 Monte Carlo simulations through April 2037.

The five chatbots’ shocks were sampled together, month by month. This preserves some of the way they moved together in the historical data.

A crossover was recorded in any month where a challenger’s monthly web visits were equal to or greater than ChatGPT’s.

The April 2037 result is one representative path in which Claude catches ChatGPT. It shows what a Claude crossover could look like under the modeled conditions. It is not an unconditional prediction.

Technical Formula Notes

Monthly change in web visits:

\Delta y_t = y_t – y_{t-1}   

AR(1) model of monthly change:

\Delta y_t = a + \phi \, \Delta y_{t-1} + \varepsilon_t

Where a is the underlying monthly change, φ is how much the previous month’s change carries into the next month, and ε is the unexpected shock in that month.

Projected web visits:

y_t = y_{t-1} + \Delta y_t

Month-over-month percentage change:

\text{MoM\%}_t = \frac{\Delta y_t}{y_{t-1}} \times 100

Residual shock:

\varepsilon_t = \Delta y_t – \left(a + \phi \, \Delta y_{t-1}\right)

Crossover condition:

y_{\text{challenger},t} \geq y_{\text{ChatGPT},t}

In plain terms, the model asks one simple question. Given how these chatbots have moved from month to month, what future paths are plausible?

Limitations

This study uses historical data only from Aug 2025 to Aug 2026. Uncertainty grows the further the forecast goes, so keep that in mind when you read anything about 2037.

Ahrefs and Semrush provide estimated web visits, not total product usage. App usage and API activity are not included.

Monthly web visits above 1 billion are rounded to one decimal place in our source data, for example, 1.2B. Percentage changes for those months are approximate.

The model assumes past behavior in web visits is a useful guide to future movement. Major launches, pricing changes, regulation, or new competitors could change the trajectory a lot.

Most importantly, the April 2037 crossover is a modeled scenario, not a prediction. It shows how Claude could catch ChatGPT, not that it will.