Newsletter · · Ashutosh Agarwal

Amazon Starts Selling Ads Inside ChatGPT Through Its DSP - Digital Ads & Retail Media Weekly - Week of September 20, 2026

Digital Ads & Retail Media Weekly for the week of September 13 to September 20, 2026. Podcast synthesis on Amazon extending its DSP into ChatGPT ad placements, new evidence that retail media budgets are mostly net new money, and a two-hour bull case that Alphabet's search ad machine is surviving AI.

Digital Ads & Retail Media Weekly

Week of September 20, 2026: Amazon Starts Selling Ads Inside ChatGPT Through Its DSP

TL;DR

  • The week's one big move: Amazon flipped a switch on September 10 that lets brands buy ads inside the ChatGPT app, bought through Amazon's own ad console and targeted with Amazon's shopping data. Two separate podcasts this week zeroed in on it, and the sharper take is that this isn't about AI at all. It's Amazon making itself the toll booth every ad dollar passes through, no matter whose screen the ad ends up on.
  • Retail media keeps proving it works. A measurement firm that has studied $45 billion of marketing spend says the returns on retail media are "really strong," the money is mostly new budget rather than stolen from Meta and Google, and it sees room for clients to spend roughly 30% more.
  • Quiet week for the usual suspects. Meta, The Trade Desk, Snap, Pinterest, AppLovin and the ad-measurement names (DoubleVerify, Integral Ad Science, LiveRamp) got essentially no serious podcast airtime. The most substantive stock discussion was a two-hour deep dive arguing Alphabet is still underrated.

What's new

1. Ads are coming to ChatGPT, and Amazon, not OpenAI, is the one selling them.

This is the story of the week, and it got covered twice from two different angles.

On Retail Media Breakfast Club, host Kiri Masters handed the mic to guest analyst Jo Lambadjieva, who wrote what Masters called the sharpest take she'd seen. Here's the plain version: as of September 10, any brand that already advertises through Amazon Ads can buy ad placements inside the ChatGPT app. The plumbing is Amazon's DSP, the "demand-side platform," which is just the buying console advertisers already use to purchase ads on Amazon's own site and, increasingly, on other people's sites too. ChatGPT is simply the newest "other people's site" on that list. Crucially, advertisers can aim those ChatGPT ads using Amazon's own shopping and browsing data.

Lambadjieva's key insight is worth quoting directly: "Amazon isn't trying to build a chatbot... It's positioning itself as the pipe that every advertising budget flows through, regardless of who owns the screen at the other end." Her point: over the past 18 months Amazon has pushed its ad business into streaming TV, premium audio, live sports, and now a leading AI assistant. Same role every time. "This isn't an AI strategy. It's an advertising aggregation strategy," she said. Why it matters: if Amazon can quietly become the cheapest, best-targeted way to buy ads everywhere, that's a slow squeeze on independent ad platforms and, eventually, on Google's share of the buying pipe.

The second angle came from operator Will Haire on Selling on Giants, who runs an Amazon agency. He confirmed the same facts (it's a US pilot, and Delta Vacations is one of the first brands testing it) and added the demand-side reason Amazon is bothering: Bain found that 24% of holiday shoppers now plan to start their product hunt inside a generative-AI tool like ChatGPT, Claude or Gemini, up from 17% a year ago. His memorable line: "The AI shelf is becoming paid media before most brands have learned how to earn placement there organically." His advice to brands was refreshingly un-hyped: don't yank money out of your best-performing Amazon search ads to chase this; give it a small test budget and measure it honestly.

2. Amazon's pitch to advertisers is "cheaper and better-targeted," and it may be working.

The reason this matters for anyone holding ad-tech stocks is the economics Lambadjieva laid out. Amazon's two-part sales pitch: price and data. On price, she cited reports that Amazon's DSP undercuts rivals on fees "sometimes dramatically," on certain deals reportedly charging close to zero versus the 15–20% that independent platforms and Google have historically taken. (She flagged these as directional figures from limited reports, not gospel, but the direction is the point.) On data, Amazon knows what you searched, compared, and left in your cart. Put cheap access to premium ad space next to that, and "budget follows gravity." She noted Amazon's ad revenue keeps growing, and "some of that growth appears to have come directly at the expense of independent buying platforms." Why it matters: this is the bear case for the open-web ad-tech names in one sentence.

3. Retail media isn't just holding up; the money is genuinely new.

On The DTC Podcast, host Eric Dick sat down with Mike Chasen of Keen, a firm that measures marketing effectiveness. ("Retail media," if you're new to the term, means the ads that retailers like Walmart, Target and Kroger sell to the brands on their shelves, a fast-growing, high-profit business now worth roughly $129 billion a year.) Chasen's headline claim: "The return on retail media has been really strong, oftentimes exceeding the ROI that you might see on other tactics." And the part investors should care about most: this spending is mostly net new; it isn't being pulled out of Meta and Google. It often comes from "trade" budgets (the money brands already spend to get good shelf placement), which sit with a company's sales team, not its marketing team. Chasen said Keen sees room for its clients to spend about 30% more on retail media than they do today, and that the fastest-growing slice is the long tail of smaller retailers, not just Amazon and Walmart. Why it matters: if retail media is additive rather than a reshuffling of existing budgets, the total ad pie is bigger than the zero-sum framing suggests.

4. Alphabet gets a two-hour bull case: search isn't dying, and the ads inside AI are already working.

On The Investor's Podcast (We Study Billionaires), investors Kyle Grieve and Shawn O'Malley made the case that Alphabet is still underrated. The ad-relevant nuggets: Google's search revenue has grown at about 14% a year over the last two years and "doesn't really show any signs of slowing down," directly contradicting the fear that AI chatbots would gut Google search. Grieve argued Google has "figured out" how to weave ads into its AI Overviews and AI Mode (the AI-written answers at the top of a search), placing ads above, below, and inside those AI answers, something ChatGPT still largely doesn't do. Why it matters: it reframes AI search as an ad opportunity for Google rather than an extinction event.

But the same two hosts made the single most useful skeptical point of the week about ads inside chatbots generally: most conversations with an AI assistant are abstract or technical and have no commercial intent. Nobody researching a niche topic wants a sneaker ad. As O'Malley put it, those chats are "really not particularly appealing to advertisers who have lots of better options" because "you don't have the same targeting at scale" as a Google search for "restaurants near me." Hold that thought; it's the crux of the debate below.

The Debate: Is AI Expanding the Ad Pie, or Quietly Draining It?

The bull case: the pie is getting bigger, and the walled gardens are getting taller.

Two forces point the same way. First, retail media is bringing genuinely new money into advertising (Chasen's point on The DTC Podcast), trade dollars that never used to count as "ad spend." Second, the companies that own first-party data and cheap inventory, Amazon and Google above all, are using AI to entrench themselves. Amazon is turning itself into the buying pipe for every screen, including ChatGPT (Retail Media Breakfast Club). Google is monetizing its AI answers better than anyone, with a billion-plus users to roll it out to (The Investor's Podcast). In this story, AI is a moat-widener for the giants.

The bear case: concentration and a shaky new ad format.

The uncomfortable flip side, also from Retail Media Breakfast Club: if Amazon becomes the toll booth for ad budgets across streaming, audio, sports and now AI chat, often undercutting everyone on fees, that's a direct threat to the independent ad-tech ecosystem, some of whose growth Amazon appears to be eating already. And the whole premise of "ads inside AI chat" rests on two things nobody has nailed yet. One, whether ads in an abstract chatbot conversation can actually be targeted well enough to be worth much; the The Investor's Podcast hosts are openly skeptical. Two, measurement: Lambadjieva argued "whoever standardizes how performance is counted on AI surfaces will effectively set the terms for anything built on top of them," and right now, no one has. Until that's solved, the "ads in ChatGPT" excitement is a promising pilot, not a proven revenue line.

The read: both sides are describing the same trend from different ends. AI is genuinely enlarging the ad market at the top (more surfaces, more shopping starting inside AI tools), while simultaneously concentrating the spoils into the two or three players who own the data and the pipe. That's great for Amazon and Google and uncomfortable for almost everyone in the middle.

Stocks in Play

Amazon (AMZN): the clear winner of the week's news.

  • Bull: Extending its DSP into ChatGPT (Retail Media Breakfast Club) makes Amazon the buying pipe for a new, fast-growing surface, on top of streaming, audio and sports. Cheap fees plus unmatched shopping data is a powerful combination, and its ad revenue keeps taking share from independents. Retail media remains a high-margin engine.
  • Bear: It's a small pilot; the AI-ads revenue is unproven and depends on measurement standards that don't exist yet. Regulators may eventually notice how much of the ad pipe one company controls.
  • Watch next: Whether ChatGPT (and other AI) ad inventory opens up beyond the pilot at lower minimum spends. Lambadjieva's advice was to "watch the access threshold, not the pilot."

Alphabet / Google (GOOGL): the contrarian long.

  • Bull: Search revenue compounding ~14% a year despite the AI scare; ads baked into AI Overviews and AI Mode; a billion-plus users to monetize; the best economics for turning AI into ad dollars (The Investor's Podcast).
  • Bear: Now Amazon is muscling into the ad-buying layer with aggressive pricing; and the giant AI/data-center spend still has to earn its keep (the hosts pegged the bar at roughly 40 cents of new annual revenue per dollar invested).
  • Watch next: Whether search-and-other revenue holds its ~14% growth as AI Overviews scale, the single number that settles the "is AI killing search ads" debate.

Meta (META): absent from the conversation.

  • No podcast this week dug into Meta's ad business. The only real read-through: an ad-financing fund manager on Strap on your Boots noted that "over 71% of businesses" now get most or all of their revenue through digital customer acquisition, overwhelmingly via Meta and Google, and that AI has automated so much of the targeting that the winning move is now "not messing up what the algorithm has already done for you." That's a backhanded bull point for Meta's Advantage+ and Google's automated tools, but it's ecosystem color, not a Meta-specific call.
  • Watch next: Q3 results and any read on Advantage+ adoption. No fresh podcast catalyst this week.

The Trade Desk (TTD): silence.

  • Zero meaningful coverage. No Kokai, no UID2, no CTV "open internet vs. walled gardens" discussion. Given the week's theme (Amazon widening its buying-pipe lead and undercutting independents on price), the absence of any TTD counter-narrative is itself a small yellow flag for the open-web bull case. Nothing new to trade on.

Read-Throughs

  • Retail media operators (WMT, CART, and the long tail): The clearest signal came from The DTC Podcast: smaller brands (under $50M in revenue) now put roughly half of their combined retail-media/trade/shopper-marketing budget into retail media, versus 15–30% for big companies, and the "everyone-else" bucket of smaller retailers is growing fastest. Instacart and Shipt's AI shopping assistants (which build a cart from a photo or recipe) came up on Selling on Giants as another sign that product discovery, and the ad inventory around it, is moving into AI-driven interfaces.
  • Emerging ad platforms (RDDT, PINS, APP, SNAP): Thin to nonexistent. Pinterest appeared only in a brand-strategy piece (The Speed of Culture, on turning "digital detox" into earned media), not an ad-revenue story. Reddit, Snap and AppLovin got no dedicated coverage.
  • Ad measurement / verification (DV, IAS, RAMP): Nothing this week. Notable given the recurring theme that measurement is the unsolved problem for AI-surface advertising; if any category should benefit from that gap, it's the verification names, yet none surfaced.
  • CTV / streaming (Roku, Disney, Netflix, Warner): No dedicated ad-inventory discussion. The one streaming-heavy podcast covered regional sports networks and content deals, not ad monetization.
  • Programmatic / adtech (CRTO, MGNI, PUBM): No standalone coverage. Criteo got a passing historical mention on a retail-media-history episode (Retail Media Breakfast Club, Retail Media's Four Eras, Sept 16); that's it.

What Changed vs. Last Week

The big ad names were quiet this week. Meta, The Trade Desk, Snap, Pinterest, AppLovin, and the entire ad-measurement group (DoubleVerify, Integral Ad Science, LiveRamp) got essentially no substantive podcast coverage.

But thin isn't the same as empty. The Amazon and OpenAI "ads in ChatGPT" launch is a genuinely important structural development, and the fact that two independent podcasts converged on it, with the smarter one reframing it as an ad-aggregation land grab rather than an AI story, tells you it's the real thing to watch. Pair that with hard evidence that retail media budgets are still growing and mostly additive, and a well-argued contrarian case that Google's search-ad machine is surviving AI just fine, and you have a coherent picture even in a quiet week: the ad market is expanding at the edges while the money concentrates toward whoever owns the data and the pipe.