The Maze: Google could lead artificial intelligence use without persuading people to abandon search. EMARKETER, the market research firm, forecasts that Google's AI Mode will pass OpenAI's ChatGPT in monthly US adoption in 2027. AI Mode lets people ask questions through Google's search service and receive conversational answers. The opportunity is familiar distribution: people can try a new way to search inside a service they already know. That is a plausible advantage, not a result the forecast proves.
The forecast puts the crossover in 2027. AI Mode reaches 41.5% of US internet users that year, up from 9.7% in 2025. It reaches 50.1% in 2030. These are percentages of internet users of any age who enter a prompt at least once per month, through a browser or app on any device. A prompt is the question or instruction someone types into the tool. The threshold measures occasional active use, not daily habit. It also differs from automatically seeing an AI summary above search results. The August 2026 forecast describes a projected future, not observed adoption through 2030.
ChatGPT keeps growing after losing first place. Its line rises throughout the period, even after AI Mode passes it. The 2030 endpoint is roughly 45.2%, estimated from the visible line position because its printed figure is hidden. That estimate indicates scale, not a recovered proprietary value. Someone can use both products during the same month. The percentages therefore cannot be added into a market-share total, and Google's projected lead does not establish that people quit ChatGPT. A bigger rival can change the ranking while both services gain users. There is no measured switch from one to the other here.
Convenience could matter as much as a better answer. Nate Elliott, EMARKETER's analyst for AI in marketing and commerce, argues that people move between conventional results and AI answers as they explore, check information and act. In that account, Google gives people both routes inside a familiar search service. The report's public overview also identifies integration into familiar tools as a driver of wider adoption. Neither the overview nor the disclosed lines isolate the effect of distribution from answer quality, pricing or other influences. Treat convenience as a credible mechanism to investigate, rather than a measured cause of the projected lead.
Monthly reach is only the start of the commercial question. A retailer could appear in an answer and still receive no visit or order. Equally, a shopper might use an answer to narrow a choice before searching or buying elsewhere. These are illustrative paths, not measured outcomes in this forecast. The practical question is which service helps relevant customers find the retailer, then whether that discovery leads to useful visits and sales. The exhibit contains no purchase, referral or advertising-return data. Its US scope also prevents a worldwide adoption claim. The full forecasting model and uncertainty ranges sit outside the readable public report, so the precise timing remains a planning assumption.
Why it matters: For businesses selling to US shoppers, the forecast argues for measuring discovery across Google and ChatGPT while keeping sales evidence separate from audience projections. It does not justify moving budgets in proportion to these percentages. Operators elsewhere can examine the same distribution mechanism, but need local usage and customer evidence before applying the US ranking. Reach earns consideration; profitable orders earn spending.
