The major AI assistants are buying more of their own advertising in second-tier US cities. Marketing Brew reported on 16 July 2026 on Sensor Tower data comparing ad impression share in the first half of 2026 against 2025, and the pattern across ChatGPT, Gemini and Claude points the same way: growth in metros like Seattle, Boston, Philadelphia, Detroit and Chicago. Every market in the dataset is a US metro. No Thai or APAC market appears in it at all. You can read the full report at Marketing Brew.
What the Sensor Tower data shows
This is third-party measurement of ad impression share, meaning where these companies' ads were served relative to everything else measured in those markets. It is not spend, and it is not a company disclosure.
| Assistant | Metro | 2025 | H1 2026 |
|---|---|---|---|
| ChatGPT | Seattle | 2.4% | 4.6% |
| ChatGPT | Boston | 2.5% | 4.1% |
| ChatGPT | Miami | 1.1% | 0.9% |
| Gemini | Seattle | 1.5% | 3.8% |
| Gemini | Philadelphia | 3.1% | 4.4% |
| Claude | Philadelphia | 2.6% | 3.2% |
| Claude | Chicago | 5.5% | 6.1% |
Claude also expanded in Detroit. The report cites the market without share figures.
One market went the other way
ChatGPT's share in Miami fell from 1.1% to 0.9% over the same comparison, despite campaigns running there. Worth holding onto, because it is the detail that stops this from being a tidy story about AI companies conquering mid-sized America.
The strategy read belongs to Sensor Tower
Sensor Tower's own interpretation is that the pattern "suggests a strategy focused on expanding adoption in less-saturated markets." That is the measurement firm's read of its own data. None of the three companies said it. The distinction matters, because impression share moving in Seattle is a fact and the reason behind it is an inference.
Marketing Brew's piece also quotes Gartner's Greg Carlucci on consumer fatigue with AI-generated advertising. "Consumer trust is an interesting piece right now," he said. It sits alongside the spend data rather than explaining it, but it works as a counterweight: more AI advertising in a market is not automatically more AI adoption in that market.
Why it matters for performance marketers
Two things, and neither is large. First, these assistants are now big enough advertisers to move impression share across whole metros, which raises auction pressure for everyone else buying in those markets. Second, if Sensor Tower's read is right, the map of where they are spending is a rough map of where they think adoption is about to grow. That would make it a leading indicator for AI-search visibility work, which pays off in proportion to how many people are asking assistants questions in the first place.
Both points depend on an interpretation the companies have not confirmed. Treat it as a weak signal.
What this means for Thai marketers
Directly: nothing. The data is US metros. There is no Thailand figure, no APAC figure, and no basis for claiming any of this describes Thai AI-assistant usage.
What can be reasoned, and this is inference rather than forecast: these platforms appear willing to spend real money buying adoption where penetration is low. Thailand is a market where penetration is low. If the same playbook reaches Southeast Asia, Thai assistant usage could step up in bursts rather than drift upward gradually, because paid acquisition compresses a timeline that word of mouth would stretch out.
That is a possibility, not a prediction, and nothing in the source supports putting a date on it. Its only practical consequence is about sequencing. If AI-assistant usage in Thailand can move quickly rather than gradually, then treating visibility inside AI answers as a project for next year carries more risk than it did last year. The work of being citable, having your entities clear and your answers structured, takes months to compound. It does not respond well to a sprint started after the demand has already arrived.
FAQ
Does this data include Thailand?
No. All markets cited are US metros.
Did OpenAI, Google or Anthropic say they were targeting smaller markets?
No. That reading is Sensor Tower's interpretation of impression-share data, not a statement from any of the three companies.
Does more AI advertising mean more AI usage?
Not necessarily. The Miami decline shows share can fall in a market with campaigns running, and Gartner's Carlucci points to consumer fatigue with AI-generated advertising as a live factor.
The one thing to take from this
US impression-share data plus a vendor's interpretation is thin ground for changing a Thai media plan, and it should not change one. The takeaway is about timing. The companies with the most information about AI assistant adoption are spending to accelerate it in under-penetrated markets, and Thailand is under-penetrated.
If that shortens your runway at all, the response is not to buy anything. It is to check whether your brand is answerable when someone asks an assistant about your category. Relevant Audience works on visibility inside ChatGPT and other assistants and on the AI search foundations that make a brand quotable in the first place.
