Google Executive Bullish on Advertising Opportunity in Agentic AI Era

Philipp Schindler, Google's chief business officer, told analysts on Alphabet's Q1 earnings call that he's "really excited about the advertising opportunities in an agentic world." Agents are software that browses and acts on your behalf. Schindler's framing: because most searches don't have one right answer, an agent sorting options leaves room to sell ads against the choices. He's presenting the death of the search box as a bigger ad market.

That framing is optimistic in ways that matter for every ad business that isn't Google.

The problem with the pitch

For Schindler's vision to pay out, three things have to be true simultaneously: agents become how most people search; Google owns that layer over OpenAI, Perplexity, and the browser makers; and advertisers pay real money for ads served to software completing tasks on a user's behalf. None of the three is proven. The third is the hardest, because people deploy agents precisely to skip the shopping process, ads included. Google is promising advertisers access to a moment the user specifically built the agent to skip.

There's also no plumbing. The entire measurement stack assumes a human clicks something. Agents don't click. No redirect, no conversion ping, no pixel fire. Quality Score proxies break. Attribution collapses quietly. Until there's an event model for "the agent finished the task," there's no ad business, only a quote.

Who gets hurt if Google figures it out anyway

The risk isn't to Alphabet's shareholders. The market swallowed "AI Overviews don't eat search revenue" quarters ago. The risk is to the independents whose pitch is built on open-web user intent.

The Trade Desk (the largest independent ad-buying platform) loses the search-adjacent intent signal it feeds into bids. LiveRamp (the leading data connectivity and identity platform) sees the attribution stack it sells devalue if intent migrates into a closed Google session. DoubleVerify (the dominant ad verification provider) loses its core product, because there's no human impression to verify. AppLovin is largely outside the blast radius; its business is mobile app advertising, not web agent flows.

The real cost isn't building a response to Google's agent ad product. It's that the signal the independent stack already runs on drains away while everyone waits. Every dollar spent on open-web identity and verification assumes intent keeps flowing through a browser.

The DOJ antitrust case against Google's existing ad business adds a layer of friction. Building a fresh ad format that could look like a new monopoly while a court is actively reviewing the old one is not a quiet move.

Our call

Google will not ship a generally available, advertiser-buyable ad format inside AI agent or AI Overview sessions with its own published measurement standard for agent completions before Alphabet's Q4 2026 earnings call in late January 2027. Medium confidence.

Schindler spoke in the future tense. An agent ad business needs an auction mechanic for a non-human recipient, a measurement event for task completion with no pixel fire, and advertiser willingness to pay CPMs against software. Building and standardizing all of that while under maximum antitrust scrutiny runs against both the engineering timeline and the legal calendar.

The near-term money stays where it is. The threat to the independents is slower than one quarter and bigger than one quote. What actually matters is whether intent migrates into a layer they can't see or measure.

Philipp Schindler, Google's chief business officer, told analysts on Alphabet's latest earnings call that he's "really excited about the advertising opportunities in an agentic world." Agents are software that browses and acts on the web for you. Schindler's bet: because most questions don't have one right answer, an agent sorting options leaves room to sell ads against the choices. He's framing the death of the search box as a bigger ad market, not a smaller one.

Here's what that means for everyone who runs an ad business that isn't Google.

What's actually being decided, and how hard to undo. This is Google staking a claim on the next query layer before a standard exists. For Google it's cheap and easy to undo: it's talk on an earnings call. For the rest of the industry it's the opposite. If agents become how people shop and Google owns the ad slot inside the agent, the open-web programmatic stack loses the one thing it sells, which is access to user intent. That's very hard to undo. The deadline isn't a date. It's whoever ships a working agent-ad format first and gets advertisers to pay against it.

The Market Analyst. Alphabet's stock already treats AI search money as a solved problem. The market swallowed "AI Overviews don't eat search revenue" quarters ago. So this comment doesn't move Alphabet much. It moves the people downstream. If Google owns the agent ad layer, The Trade Desk loses the search-adjacent intent signal it feeds into bids, and DoubleVerify loses the thing it verifies, because there's no human impression to check. AppLovin is mostly out of the blast radius: its money is in mobile apps, not web agent flows. In plain terms: the risk here isn't to Google's shareholders, it's to the independents whose whole pitch is "we see user intent on the open web."

The Skeptic. An executive being "excited" about revenue that does not exist yet is not a plan. For agent ads to pay, three things all have to be true at once. Agents have to become how most people search. Google has to own that layer. And advertisers have to pay real prices for ads shown to software, not people. Each one is a stretch, and users deploy agents precisely to skip ads. Add the live DOJ antitrust case against Google's ad business: building a fresh ad monopoly on top of the one a court is dismantling is not a quiet move. In plain terms: this is a confident answer to an analyst, built on a surface nobody has shipped.

The Operator. Try to act on this Tuesday morning and you hit a wall. Ask your Google rep what the auction looks like when an agent, not a person, receives the ad. There's no answer yet. The 90-day problem: your campaign managers keep optimizing for clicks and pixel fires that agents don't produce. The agent completes the task inside its own session. No redirect, no conversion ping, no pixel. Quality Score proxies break. Attribution quietly collapses. Paid search teams feel it first; programmatic display teams feel it when budgets drift toward whatever Google's agent format turns out to be. In plain terms: the measurement plumbing everyone runs assumes a human clicks something, and agents don't click.

The Customer / End User. Two customers here, and they want opposite things. The advertiser wants to reach a buyer and will follow intent wherever it goes, including into an agent, if someone can prove the ad worked. The consumer running the agent wants the opposite. People hand a task to an agent to avoid the shopping slog, which includes the ads. In plain terms: Google is promising advertisers access to a moment the user specifically built the agent to skip. That tension doesn't resolve with a nice ad format. It resolves with whoever controls the agent deciding what the user sees.

The CFO. The cost that matters isn't building an agent ad product. It's the signal your current business runs on draining away while you wait. If intent moves into a closed Google agent layer, every dollar the independents spend on open-web identity, attribution, and verification is spent chasing a surface that's shrinking. The payback math on The Trade Desk's and LiveRamp's current stack assumes intent keeps flowing through a browser. In plain terms: the real risk is that the asset you already paid for quietly stops being worth what you paid.

The tensions. The real disagreements:

1. Is this a product or a press release? The Strategist read in this window calls Schindler's framing the smartest move Google's made in a decade, redefining a threat as expansion. The Skeptic calls it a holding-pattern answer. Both can't be right. The difference is whether Google ships an agent-ad format advertisers actually pay for, or keeps saying "excited" until a court or a competitor forces the issue.

2. Who gets disintermediated, and when? The Market Analyst sees a slow bleed over 2 to 3 quarters. The Strategist sees the open-web stack losing its leverage point entirely. The gap is timing: a gradual repricing versus a structural break.

3. Does consumer behavior even cooperate? Everyone assumes agents become the dominant way people search. The Customer lens says the people adopting agents fastest are the ones most hostile to being advertised to inside them.

What it hinges on. Three beliefs, and the whole story turns on them. First, that agents become how most people search, not a power-user toy. Second, that Google owns that layer rather than OpenAI, Perplexity, or a browser maker. Third, that advertisers pay real money for ads served to software. Right now the council leans skeptical on the near term and nervous on the long term. Skeptical because none of the three is proven and the DOJ case cuts against Google building anything that looks like a new monopoly. Nervous because if all three land, the independent stack has no answer. What to de-risk before believing the TAM-expansion story: watch for an actual auction mechanic and a measurement standard for agent completions. Until there's an event model for "the agent finished the task," there is no ad business, only a quote.

Prediction: Google will not ship a generally available, advertiser-buyable ad format inside AI agent or AI Overview sessions with its own published measurement standard for agent completions before Alphabet's Q4 2026 earnings call in late January 2027.

Confidence: Medium. No event model exists for agent completions, and the DOJ remedy fight makes a new ad monopoly a bad look right now.

Why: Schindler's comment is an earnings-call framing, not a product announcement, and he described the opportunity in the future tense ("we can actually provide"). An agent ad business needs three things that don't exist yet: an auction mechanic for a non-human recipient, a measurement event for a task the agent completes inside its own session with no pixel fire, and advertiser willingness to pay CPMs against software. Building and standardizing all of that in under a quarter, while the DOJ is actively trying to break up Google's existing ad stack, runs against both the engineering timeline and the legal caution. The opposite outcome, a shipped buyable format with published measurement this fast, would require Google to move faster on agent ads than it has on any ad surface in years and to do it under maximum antitrust scrutiny.

Revisit by 2027-02-15: We're right if, by Alphabet's Q4 2026 earnings call, there is no generally available ad product any advertiser can buy against agent or AI Overview sessions with a Google-published measurement standard for agent completions. We're wrong if Google launches such a buyable format with a stated completion-measurement spec before that call.

The near-term money stays where it is. The thing that actually threatens the independents is slower than one quarter and bigger than one quote, and it's whether intent migrates into a layer they can't see.