Read the past week of trade coverage on AI advertising in isolation and each story sounds like the same shift: marketing is moving toward chatbots and AI-mediated search. Read the stories together and a different picture appears. Three separate bets are being placed at once, on three different layers of the stack, and nobody covering them has connected the three into a single question: does any of this add up to a channel a marketer can actually plan a budget against yet?
Three accounts, three layers
AdExchanger’s reporting on ChatGPT advertising treated the product as already real but not yet investable. In “ChatGPT Ads Are Here. Now Comes the Hard Part,” the outlet quoted Mary Gabrielyan, chief strategy officer at AI Digital, saying that while cost-per-click bidding, pixel tracking and conversion APIs make the ad unit accessible with minimal spend, “it’s hard to call this a meaningful media channel yet” without a way to trace ChatGPT’s influence on cross-channel performance or tie it to offline outcomes. AdExchanger’s framing is a measurement story: the ad slot exists, the accountability layer around it does not.
Digiday reported a layer beneath that. In “OpenAI is coming for SMB advertisers,” it found OpenAI recruiting for six open roles, spanning data science, growth and mid-market demand strategy, to build a dedicated small-and-midsize-business ad unit, pulling talent from Meta, Google, LinkedIn, TikTok, Amazon and Reddit. Digiday’s own reporting states plainly that the data infrastructure for that business “doesn’t exist yet.” eMarketer principal analyst Nate Elliott told the outlet that SMBs “make up the majority of ad revenue for Google and Meta” and that “any slightly ambitious ad business” has to go after them eventually. Digiday’s framing is an infrastructure story: OpenAI is staffing up for a self-serve SMB ad business before the plumbing for it is built.
Marketing Dive covered a third layer entirely, upstream of any ad unit. In “Behind Reddit and YouTube’s roles in AI visibility,” it reported that YouTube videos are 4.3 times more likely to surface in Google’s AI Overviews than in standard search, while ChatGPT cites Reddit 2.4% of the time, more than twice as often as YouTube’s 0.99%. Reddit’s ad revenue grew 64% year over year to $762 million in its most recent quarter. Marketing Dive’s framing is a content-supply story: before a marketer buys an ad slot inside an AI answer, the more consequential fight is which platform’s content the AI model was trained on and chooses to cite in the first place.
Where the three accounts disagree
The point of friction is direct. AdExchanger’s sourcing treats ChatGPT advertising as a live, testable unit today, with a real cost-per-click mechanism that campaigns are already running against. Digiday’s sourcing, reported in roughly the same week, describes the commercial infrastructure behind that same advertising business as not yet built. Both cannot be fully right about the same product. The more accurate synthesis is that OpenAI has shipped a self-serve ad interface faster than it has built the sales, data and measurement operation to support it at SMB scale, which is exactly the gap AI Digital’s Gabrielyan is describing from the buyer’s side. The ad unit is ahead of the ad business. Neither trade outlet stated that combined conclusion; each only had its own half of it.
The disagreement matters because it maps to a pattern search advertising went through two decades ago and largely resolved: a self-serve interface shipped years before Google built the account management, agency co-op programs and attribution tooling that eventually made search a defensible line item for a small business owner. Digiday’s detail that OpenAI plans to lean on outsourced vendors for SMB sales, mirroring how Google and Meta built their own SMB channels, is the clearest evidence that OpenAI is following that same playbook deliberately rather than improvising it. The difference this time is speed: search advertising took years to build its SMB sales motion after launch. OpenAI is trying to compress that timeline while the ad unit is already live and being tested by early buyers, which is exactly the sequencing AdExchanger’s sourcing is complaining about from the demand side.
What it means for the marketing leader
Treating “AI visibility” as a single new budget line is the mistake this coverage collectively exposes. It is at least three separate wagers, on three different clocks:
The ad-buying wager
ChatGPT’s ad unit is accessible now, at low minimum spend, but without the attribution tooling that makes a paid-search or paid-social line item defensible in a budget review. Treat early spend here as a test budget, not a scaled channel, until measurement catches up to the ad unit itself.
The infrastructure wager
OpenAI’s SMB hiring push signals where the company expects volume to eventually come from, not where it is today. A self-serve interface that works for an enterprise pilot is not the same product as one with the demand-side tooling, agency relationships and support layer that SMB advertising requires at scale. Watch OpenAI’s hiring and vendor partnerships, not just its ad-manager UI, for the real signal of readiness.
The content-supply wager
Which AI system cites which platform is already being decided, upstream of any ad product, by what each model was trained on and what it ranks as authoritative. Marketing Dive’s figures show that gap is not marginal: a four-to-one citation advantage for YouTube inside Google’s own AI Overviews versus a Reddit citation rate in ChatGPT that is more than double YouTube’s, because the two platforms have built structurally different relationships with the two model owners. That is a battle brands are already fighting without a media budget, since no ad unit currently lets a marketer buy their way into a model’s training data or citation preference. Zoom’s marketing team, for one, has bet on it directly. The company partnered with journalist and podcast host Nayeema Raza to build a slate of interviews and a limited podcast series specifically to earn the kind of third-party authority large language models weight when deciding what to surface. “Ultimately, our job is now to be found, to be present, to be believed, and to be preferred,” Zoom chief marketing officer Kimberly Storin said of the strategy in the company’s announcement. That is a bet on content and credibility, not on an ad unit at all, and it is happening in parallel with the ad-buying and infrastructure wagers, not after them, using a budget line most finance teams still categorize as public relations or influencer spend rather than performance media.
The three accounts do not describe one emerging channel. They describe three different races, on three different timelines, run by different teams inside the same set of companies, with no shared measurement standard tying the outcomes of one to the others. A brand can win the content-citation race and still have nothing to show for it in a paid-media report. It can spend against the ChatGPT ad unit and still have no way to prove it moved anything. Marketing leaders evaluating “AI visibility” spend this quarter should ask which of the three races a given budget line is actually entered in, because right now the industry’s own coverage of it cannot tell the difference and neither, per MarTech’s own reporting on the shared measurement standard effort, can the standards bodies trying to build a common yardstick. The gap between what platforms are citing and what search still ranks, which MarTech covered in its report on AI Overviews citing a different web than search ranks, is the same structural problem one layer up: measurement has not caught up to any part of this stack yet, at the ad level, the infrastructure level, or the content level.
Source: Zoom Newsroom
