AI broke search into four ecosystems.

The search line on a 2026 media plan now covers four different behaviors. Our State of Digital Media 2026 report tracks discovery splitting into four ecosystems that brands must address separately: generative AI, vertical search, social-as-search, and voice and visual (MediaPath Research, The State of Digital Media 2026, May 2026). Each has its own platforms, its own ad surfaces, and its own measurement posture. A single search budget, planned and reported as one channel, no longer describes where discovery actually happens.

The money is already moving. eMarketer forecasts AI search ad spend rising from $2.08 billion in 2026 to $25.93 billion by 2029 (eMarketer, 2025). Forecasts that steep are directional by nature, and this one has unusually hard usage numbers underneath it: the platforms involved now count their users in the hundreds of millions to billions, and two of them, Alphabet on AI Overviews and Microsoft on Copilot seats, disclosed in earnings filings.

This piece covers four parts of the shift: the platform scale table and the disclosures behind it, the contraction in the organic click economy underneath it, how the youngest buyers already move across the four ecosystems, and a planning posture that treats discovery as four budgets and four measurement postures. Figures are drawn from The State of Digital Media 2026 and carry their issuer where one is named.

OpenAI disclosed that ChatGPT crossed 900 million weekly active users in February 2026, double the year before, alongside 50 million paid subscribers and more than $25 billion in annualized recurring revenue (OpenAI blog, February 27 2026). Google answered at I/O in May 2026: the Gemini app reached 900 million monthly active users and AI Mode hit 1 billion, while Alphabet's Q1 2026 earnings put AI Overviews above 2.5 billion monthly users (Google I/O 2026; Alphabet Q1 2026 earnings).

The report's channel scorecard reads the race as Google AI holding a 60%-plus share of AI search usage with OpenAI above 25%, reported as floors rather than exact shares, concentrated at the top and still contested underneath (MediaPath Research, The State of Digital Media 2026, May 2026). Microsoft counts 15 million paid Copilot M365 seats and has announced Sponsored Answers, which puts a fifth ad surface in the table (Microsoft FY26 Q2, January 28 2026).

Every row of the table carries an ad surface that is live or announced: sponsored placements inside AI summaries, conversational placements in AI Mode, sponsored answers and ChatGPT Shopping at OpenAI, in-conversation native placements in Gemini, and Microsoft's announced Sponsored Answers in Copilot. The reach is mainstream, and the monetization layer is now being fitted on top of it. That combination is what moves AI search from a watching brief to a plannable line.

The organic side of search is giving ground faster than the paid side is forming. Ahrefs measured Google AI Overviews cutting position-1 organic click-through rates by 58% as of December 2025, up from a 34.5% cut in April 2025, at exactly the position most SEO strategies are built to win (Ahrefs AI Overviews study, February 2026). Our report reads that curve as accelerating (MediaPath Research, The State of Digital Media 2026, May 2026).

The planning implication is concrete. Organic benchmarks set before 2025 no longer describe the baseline, because position-1 organic click-through is now about 42% of what it was (Ahrefs AI Overviews study, February 2026). The answers are increasingly consumed where they are generated, and the click that used to mark a discovery does not always happen. For brands that is a visibility question no analytics dashboard was built to answer.

Presence inside the answers themselves runs through the paid surfaces in the table above, and those surfaces are the earliest and least standardized part of the stack, which argues for testing them while the formats are still forming, and for reading every organic report against the new baseline.

More than 80% of Gen Z uses AI tools, per the report's channel scorecard, which flags a broad trust gap alongside that adoption (MediaPath Research, The State of Digital Media 2026, May 2026). The same scorecard reads AI search as a discovery overlay on existing search instead of a separate time pool, which is why its influence on what gets found outruns its share of measured hours.

Social already functions as search for the same cohort. More than 80% of Gen Z uses social platforms as search, inside roughly 2 hours 18 minutes of daily US adult social time, with Meta, YouTube, and TikTok holding about 70% of it (eMarketer, 2026; MediaPath Research, The State of Digital Media 2026, May 2026). A product search that starts on TikTok never touches a results page, and a brand that only monitors the results page never sees it.

Vertical search and voice round out the four. The report names both as ecosystems brands must address separately, and flags voice search and conversational AI as the AI-native behaviors of the youngest cohort, without sizing either as a search channel (MediaPath Research, The State of Digital Media 2026, May 2026). Each of the four is a separate place a brand can be present or absent, and the youngest buyers already move across all of them.

Give each ecosystem its own line. A generative AI line funds early tests on the live surfaces (sponsored placements in AI summaries, sponsored answers, ChatGPT Shopping) while formats and pricing are still forming. A vertical search line covers retail platform discovery, where the audience arrives purchase-qualified. A social-as-search line briefs creative for queries that never reach a search box. A voice and visual line stays small and instrumented until the ad surfaces catch up with the behavior.

Change the measurement with the split. A growing share of discovery now happens inside answers rather than through a click path, and identity fragmentation already broke last-touch attribution. The report's broader measurement finding points the same direction: 46.9% of US marketers rate MMM the most reliable cross-channel measure, while last-click MTA holds only a 19.4% reliability share, because model-based measurement is what compares cleanly across surfaces that do not share a click (eMarketer/TransUnion, July 2025).

Move while the formats are forming. The unit of discovery planning is now the behavior, not the search box. Budget the four ecosystems separately, test the generative surfaces before formats and pricing standardize, and benchmark organic search against the post-2025 baseline, which the CTR data has already replaced. eMarketer's forecast curve runs from $2.08 billion in 2026 to $25.93 billion in 2029, and the front of that curve is where the learning happens (eMarketer, 2025).