Meta's new Muse models have revived a sharper version of the threat to Google Search. The issue is not whether an AI can answer a question. It is whether an agent can complete the commercial task that used to send a user through Google's search results.
Deepwater Asset Management's Gene Munster and Doug Clinton argued that Muse could pose a larger long-term risk to Alphabet than conventional chatbots. Their reasoning is that many early chatbot questions were not heavily monetized by search, while an agent that carries out multi-step work can move closer to purchases, bookings and other high-value actions. That is an interpretation, not evidence that advertising revenue has already shifted.
Meta's own announcement provides the product foundation for the thesis. Muse Wonder is available through Meta's API and Muse Code, while Muse Wonder Full is available to Muse Code subscribers. Meta said the standard model will also power its Muse assistant. The company described the models as capable of complex reasoning, coding and agentic workflows, but it did not disclose Muse revenue, usage at scale or transaction conversion data.
Google also has powerful defenses: a global search habit, advertiser relationships and the ability to integrate agent functions into products users already visit. The competitive question is therefore distribution as much as model quality. Meta can place an assistant across large consumer platforms, while Google can connect AI actions to Search, Gemini and its advertising system.
Investors should treat the Muse launch as a credible strategic challenge, not a completed transfer of economics. Evidence would include rising Muse usage, successful task completion, merchant participation and new revenue that does not simply displace activity elsewhere inside Meta. For Alphabet, the relevant signal is whether commercial queries migrate away from search or remain inside Google's own AI interfaces. Until those metrics appear, the risk is real but still prospective.