Newsletter · 2026-08-10 · Ashutosh Agarwal

Retail Media Becomes a Permanent Tax on Selling Online - Weekly Online Ads Podcast Recap - Week of August 10, 2026

For the week of August 3 to 10, 2026, eighteen episodes made the same case: Amazon's ad business grew 26% to $19.8 billion, a full 10 points faster than the third-party seller business beneath it, while answer engines began turning into a genuine ad channel.

Weekly Online Ads Podcast Recap

Week of August 10, 2026: Retail Media Becomes a Permanent Tax on Selling Online

Covering podcasts published 2026-08-03 through 2026-08-10. Eighteen relevant episodes surfaced inside the 7-day window, so no older "fallback" coverage was needed.

Executive Summary

The retail-media machine keeps outrunning the store. Amazon's ad business grew 26% last quarter to $19.8 billion, a full 10 points faster than the third-party seller business it sits on top of. The takeaway from operators: paying to be seen is now a permanent, growing tax on selling online, not a holiday-season blip (Selling on Giants).
The Trade Desk (TTD) had a genuinely bad week and became the sector's cautionary tale. Growth collapsed from 12% to 3% and management guided to a possible ~12% revenue decline next quarter; the stock fell about 20% on the print and is down roughly 90% from its high. The debate: is this a company-specific execution failure, or proof the "open internet" is losing to Amazon, Google and Meta? (Motley Fool Hidden Gems)
Meta's ad engine is firing (28% growth) but its story is a mess. Commentators say the core business is "absolutely crushing it," yet the market punished the stock because management can't explain what its enormous AI spending will actually deliver, a stark contrast to how clearly Microsoft and Amazon narrated their cloud economics (The Investing for Beginners Podcast).
AI chatbots are quietly becoming ad platforms. ChatGPT is pulling in display dollars (not search dollars), is launching Criteo-style product-carousel ads for the holidays, and is partnering with AppsFlyer on measurement, a more open, data-sharing posture than Google's walled approach (Marketecture).
Generative-AI ad creation moved "from fear to confidence." Adobe's Firefly is now used by the NFL, L'Oréal and Disney, and Adobe just shipped a ChatGPT plugin tapping 70+ of its tools, while a Canva survey found 70% of consumers still think AI-made ads are "missing their soul" (Marketing Vanguard; Everyday AI; Inside Insights).
AppLovin (APP) is the split-screen name of the week: performance marketers say the product genuinely works, while analysts warn the stock is "trading on momentum" with a gap between the product's value and the share price (Marketecture; DTC Podcast).
YouTube is now treated as real TV, and money is flowing toward "performance TV" on connected screens, though experts warn that judging TV by search-style return targets quietly destroys a brand (Next in Media; Marketecture).

(a) Dominant Themes

1. Retail media is now a structural tax on selling: and it is still accelerating

The single most-repeated data point of the week was Amazon's advertising growth. On Selling on Giants (Aug 4), host Will Haire of Bellavix laid out Amazon's Q2 line by line: total sales of $200.6 billion (up 20%), third-party seller services of roughly $46.8 billion (up 16%), and advertising of $19.8 billion (up 26%). His point is that the ad business is compounding about 10 points faster than the selling business it rides on:

"Advertising is growing 10 percentage points faster than the seller services revenue it sits on top of… This is not a temporary holiday bidding spike that disappears when everyone calms down in January. It is part of the business model."

He put the warning in plain (and funny) terms for any brand still modeling flat ad costs:

"If your 2027 plan assumes your ACoS stays flat while Amazon's advertising business keeps compounding faster than marketplace services, the spreadsheet is doing motivational speaking. It is not forecasting."

Haire also flagged where that growth goes next, into AI-driven shopping. Shoppers who click a sponsored prompt "convert 48% more often and spend 21% more"; Amazon's automated ads agent is producing an "8% lower cost per impression and a 6% lower cost per acquisition"; and Alexa-based shopping has crossed 350 million users in twelve months, with interactions up more than 5x year over year and US users spending about 40% more per order. His advice to sellers: model future margins with ads taking a bigger bite, and clean up product data because "when software helps choose the product, structured data becomes part of your creative strategy." He also noted that Walmart "finally gave advertisers a tool Amazon sellers have used for years."

The theme carried into back-to-school planning on Better Advertising with BTR Media (Aug 4), where BTR's Gibby and Laura described a softening consumer forcing more precise spending across Amazon, Walmart and Target retail-media networks. A striking detail: "62% of shoppers have already started their journey" before the school-bell rush, and Walmart's new feature that lets parents select their school online and auto-fill a cart is quietly reshaping product discovery. Their guidance was to spread budgets "a little bit thinner and direct them towards a more appropriate audience", leaning on last year's purchasers and tight audience segments rather than broad in-market targeting.

2. AI has become the operating system of advertising: on both the creation and the delivery side

Two things are happening at once. First, AI is now writing and sizing the ads. At Cannes, Adobe CMO Lara Balazs told Marketing Vanguard (Aug 4) the industry mood has shifted "from fear to confidence," reframing the story as "AI augmenting humans" rather than replacing them. Her proof points were real brands on Adobe's Firefly studio, the NFL, L'Oréal, and a Disney "Imagineering" partnership using Adobe Firefly Foundry, a proprietary model trained only on Disney's own IP so it stays "commercially safe." She argued the value is grinding out the drudgery, not the ideas: Adobe's own Cannes activation required 150 static assets, 170 films and "200 plus aspect ratios," and "no creative wakes up and says, I want to size assets all day."

That same week, Everyday AI (Aug 7) reported Adobe shipping a unified plugin inside ChatGPT that reaches "more than 70 tools from Photoshop, Firefly, Acrobat, Express, Premiere, Lightroom, Illustrator, InDesign and Stock", you describe the job and it picks the right Adobe tool. Host Jordan Wilson noted Adobe is "running this play on every chatbot," with tools already live on Claude and Copilot, but called the ChatGPT version "the first one we've seen that's this robust."

Second, AI is deciding what ad you see. On The Investing for Beginners Podcast (Aug 6), guest Thomas Chua explained that Meta used to recommend Reels mostly from the caption and hashtags a creator wrote, but now a large language model "is reading through every single one of the content… and is able to better allocate this content to the users," and "likewise, they're deploying it at their advertising side also." That is the mechanism behind Meta's re-accelerating ad growth.

3. The "answer engine" is turning into an ad channel

The most forward-looking discussion came from Forrester analyst Nikhil Lai on Marketecture (Aug 7). His nuanced read is that ChatGPT is not yet stealing Google's core search money:

"The money being spent on ads are mostly display dollars… rather than search dollars. Paid search remains fully funded and display dollars that are chasing higher yield and chasing greater attention are going to ChatGPT."

He noted OpenAI is launching "product carousel ads… these like catalog listing ads that come directly from your product set" for Q4, "a very Criteo like format", and is partnering with AppsFlyer "to bring in-app measurement to ChatGPT ads." His bigger point: OpenAI is being "a lot more flexible on data sharing" than the Google model of controlling everything top to bottom. He also mentioned Reddit's ad business is "up 64%," a standout growth rate.

4. Measurement and "performance TV" are the industry's summer obsession

Lai called incrementality and attribution "the topic of the summer," now being "operationalized… always-on" at the most advanced brands. The connected-TV land grab was everywhere: on Next in Media (Aug 6), Bright Mountain's Andy Davidson and Deep Focus's Kyle Krueger described YouTube as "the undisputed number one streaming platform," with indie agencies now seeing "almost a parity with what you would receive from traditional TV" thanks to its targetability and measurability. They also described using "synthetic" and "agentic" audience personas to adjust TV buys "in almost real time."

(b) Active Debates

Debate 1: The Trade Desk: execution stumble, or is the open internet losing?

This was the sharpest disagreement of the week, on Motley Fool Hidden Gems (Aug 7). The facts: The Trade Desk's growth fell from 12% in Q1 to just 3% this quarter, with guidance for a possible "12% drop in revenue in the upcoming quarter," and the stock down ~20% on the report and "about 90% from their high," now at roughly a $5 billion enterprise value and a 6.6x forward P/E.

"It's execution" (bearish on the excuse): One host noted CEO Jeff Green spent "10 minutes" on the call blaming macro, "Nike isn't spending more and all these kinds of things", and called it a red flag: "This doesn't pass the sniff test." His evidence was that rivals in the same space are growing fine: "Magnite and Pubmatic reporting double digit growth. Even companies like Zeta is posting very strong growth rates." Verdict: internal execution problem, and Green "needs to move upstairs" to executive chairman because "his voice is no longer resonating on Wall Street."
"The walled garden is winning" (structural bear): Another host argued the platforms are simply taking the share: "I think the walled garden is winning… the internet is getting more closed off. The trade desk has always said we have a solution for that. We're not seeing it." He added a genuinely new worry, as AI bots begin doing shopping, "Do bots respond to advertising?"
Value or value trap? Both hosts landed on "value trap for now," warning the cheap-looking multiple will rise as revenue declines and margins compress. "Fool me once, shame on you. Fool me twice, shame on me."

Debate 2: Meta: great ad business, terrible AI story

On The Investing for Beginners Podcast (Aug 6), Thomas Chua drew a sharp line between Meta's operations and its communication. On the business: "The ad business grew 28%, impression growth, engagement growth, both still double digit, despite being such a mature business… they are absolutely crushing it on their core business." On the stock: the market rewarded Microsoft and Amazon but "punished Meta severely" because Satya Nadella and Andy Jassy "were very, very clear in communicating about the economics" of their cloud businesses, while Meta left a question mark over what its Superintelligence Lab spending will return. His verdict on Zuckerberg's choice to stay vertically integrated (rather than sell compute as a hyperscaler): "I kind of think that that's the right move, just very, very poorly communicated." He tied the strategy back to 2022, when Apple's iOS 14 privacy change "severely impeded Meta's business", leaving Zuckerberg determined never again to be "held hostage" by another company's platform.

Debate 3: AppLovin: does the product justify the stock?

Two very different lenses. On Marketecture (Aug 7), Forrester's Nikhil Lai separated the company from the share price: "There is a disassociation between the intrinsic value of the product and the company and the extrinsic value of the stock… it's trading on momentum." Yet he confirmed the product delivers, "for the most part folks are finding consistency with it." From the buyer's seat, the DTC Podcast (Aug 7) team was enthusiastic but candid about the cost of using it: "AppLovin… we love it. We do very well there but it's like it requires so much volume" of bespoke creative per channel. On Squawk on the Street (Aug 6), Jim Cramer noted only that investors treat AppLovin as a company that "never misses."

Debate 4: Does AI-made advertising work, or does it lack a soul?

Inside Insights (Aug 3) opened with a telling statistic from Canva's 2026 State of Marketing and AI report: "70% of consumers say AI-generated advertising feels like it's missing its soul, even as 99% of marketers plan to increase their AI investments this year." Canva's Matt Harris and Oli Bustle described squaring that circle across five European campaigns: they used AI for survey analysis and near-instant storyboards to "narrow the gap between what the creatives want to put across and what the consumers see," but deliberately made their Italy ad in painstaking lo-fi stop-motion, "the most lo-fi, unautomated, painstakingly slow process you can imagine", to humanize a product that is itself an AI tool. Their framing: "we're not using AI in our ads, but we are ultimately advertising AI." This is the counterweight to Adobe's confident Cannes narrative, the tooling is ready, but consumer trust is not automatic.

Debate 5: Is "performance TV" a smart evolution or a trap?

On Marketecture (Aug 7), Nikhil Lai split connected-TV buyers into two camps: legacy linear budgets migrating to CTV to replace lost reach, versus "new-to-TV" performance brands treating TV "like next gen search or next gen social." He argued the second group is making a mistake: "I am bearish on CPA guarantees on TV… if you begin to try to hold TV accountable to a really aggressive [performance] goal, you are going to lose share of voice in the process," and "brands don't grow by optimizing for the lowest possible CPM or the lowest possible CAC." He noted the Walmart–Vibe deal has now closed (Vibe is part of Walmart) and Pinterest is pushing into the performance-TV space, with "TV Scientific" also in the mix.

Debate 6: Is Google's AI leadership shakeup a crisis for the ad giant?

Not an ad debate strictly, but relevant to the company behind Search and YouTube. The AI Daily Brief (Aug 6) covered DeepMind CEO Demis Hassabis stepping aside (staying as chairman and becoming chief scientist for all of Google) and legendary engineer Jeff Dean leaving to found an independent automated-research company, Discovery Loop. The bear read is a continuing "brain drain", John Jumper left for Anthropic, Noam Shazir for OpenAI, with the market marking Google down about 4% on the day. The counterpoint offered on the show: this may simply formalize a reality in which Hassabis had "already been checked out for the better part of a year," freeing him to focus on frontier AGI work. Either way, it lands against the backdrop of the AI-Overviews-versus-Search-revenue question that hangs over Google's ad franchise.

(c) Tickers Mentioned

AMZN (Amazon), retail media / Alexa commerce

Bull: Advertising is the fastest-growing, highest-margin part of the flywheel; AI shopping (Alexa, ads agent) is extending it. Bear (implied): Rising ad load is a growing cost burden on the sellers who fund it.
Source: Selling on Giants, "Amazon's Ad Tax, Seller Financing & Walmart Ad Controls", Will Haire (Bellavix), 2026-08-04.
Quote: "Amazon Sales grew 20% last quarter. Amazon Advertising grew 26%. That six-point gap may be the most important number in Marketplace Commerce this week." Also cited on Marketecture (Nikhil Lai, Forrester, 2026-08-07): "almost a $20 billion quarter for Amazon ads up 26% year over year, stock went crazy."

WMT (Walmart), Walmart Connect / Vibe acquisition

Bull: Closing the retail-media feature gap with Amazon; back-to-school cart-automation and the closed Vibe deal push it into performance TV.
Source: Selling on Giants (Will Haire, 2026-08-04); Better Advertising with BTR Media (Gibby & Laura, 2026-08-04); Marketecture (Nikhil Lai, 2026-08-07).
Quote: "Walmart finally gave advertisers a tool Amazon sellers have used for years." (Selling on Giants)

META (Meta Platforms), core ads, AI ad delivery, capex story

Bull: 28% ad growth with double-digit impression and engagement gains; LLMs now read all content to target Reels and ads; "the strongest advertising player with the strongest growth right now." Bear: Superintelligence-Lab spending is a question mark and was "very poorly communicated," so the market punished the stock.
Source: The Investing for Beginners Podcast, "The AI CAPEX Illusion with Thomas Chua," 2026-08-06.
Quote: "The ad business grew 28%, impression growth, engagement growth, both still double digit, despite being such a mature business… they are absolutely crushing it on their core business."

GOOGL / GOOG (Alphabet), Search/YouTube ad engine, AI leadership

Bull: Core Search, YouTube, AI Overviews and Gemini can absorb excess AI compute; ChatGPT is so far taking display, not paid-search, dollars. Bear: Leadership "brain drain" (Hassabis stepping back, Dean leaving) and open questions about AI Overviews' effect on Search revenue; stock down ~4% on the shakeup.
Source: The AI Daily Brief, "Google's AI Leadership Shakeup," 2026-08-06; Marketecture (Nikhil Lai, 2026-08-07).
Quote: "Paid search remains fully funded and display dollars that are chasing higher yield and chasing greater attention are going to ChatGPT." (Marketecture)

TTD (The Trade Desk), independent DSP / open internet

Bull (fading): Cheap at ~6.6x forward earnings if the strategy is right and headwinds are macro. Bear: Growth cratered 12% → 3% with a possible ~12% revenue decline guided; "the walled garden is winning"; leadership credibility gone; "value trap for now."
Source: Motley Fool Hidden Gems, "The Trade Desk's Woes & A New AI Donut?", 2026-08-07.
Quote: "12% growth in the first quarter, only 3% growth in this quarter and forecasting potentially a 12% drop in revenue in the upcoming quarter… This is an execution thing."

MGNI (Magnite) & PUBM (PubMatic), ad-tech / SSPs

Bull: Used as the counter-evidence against The Trade Desk, both "reporting double digit growth," suggesting the weakness is TTD-specific, not sector-wide.
Source: Motley Fool Hidden Gems, 2026-08-07.
Quote: "Magnite and Pubmatic reporting double digit growth. Even companies like Zeta is posting very strong growth rates."

ZETA (Zeta Global), marketing/data platform

Bull: Named as a growth alternative to The Trade Desk and a position one host has been "adding to."
Source: Motley Fool Hidden Gems, 2026-08-07.
Quote: "One that I have been interested in recently and been adding to my position too is Zeta Global."

CRTO (Criteo), commerce/retail-media ad tech

Bull: Positioned as the template for where AI-chatbot commerce ads are heading (product-carousel/catalog formats); Criteo's own CEO seeing more upper-funnel searches in AI answers.
Source: Marketecture, 2026-08-07.
Quote: "They're launching product carousel ads… which again is a very Criteo like format."

APP (AppLovin), mobile ad platform / AXON

Bull: Product works with consistent incrementality; performance marketers "do very well" on it; briefly noted hitting an Axon platform milestone and expanding into e-commerce. Bear: Stock is "trading on momentum" with a "disassociation" between product value and share price; success demands enormous volumes of channel-specific creative.
Source: Marketecture (Nikhil Lai, 2026-08-07); DTC Podcast (2026-08-07); Schwab Network (2026-08-05); Squawk on the Street (2026-08-06).
Quote: "There is a disassociation between the intrinsic value of the product and the company and the extrinsic value of the stock… it's trading on momentum." (Marketecture)

RDDT (Reddit), ad ramp

Bull: Standout growth, advertising cited as "up 64%."
Source: Marketecture, 2026-08-07.
Quote: "Reddit ads up 64% to give you some context."

PINS (Pinterest), performance TV expansion

Bull: Entering the performance-TV / connected-TV market, generating industry "chatter."
Source: Marketecture, 2026-08-07.
Quote: "Pinterest entering this market… the Walmart vibe deal has gotten more chatter."

ADBE (Adobe), generative-AI ad creative

Bull: Firefly adopted by the NFL, L'Oréal and Disney (via IP-protected Firefly Foundry); new ChatGPT plugin exposes 70+ Adobe tools; running the same play across Claude and Copilot.
Source: Marketing Vanguard (Lara Balazs, CMO of Adobe, 2026-08-04); Everyday AI (Jordan Wilson, 2026-08-07).
Quote: "With Disney, they have an Imagineering partnership with us where they're using Adobe Firefly Foundry, our proprietary model that only trains on their IP." (Marketing Vanguard)

Also referenced: DIS (Disney), as a marquee Firefly Foundry customer (Marketing Vanguard) and in earnings context alongside a TikTok streaming discussion on the Elon Musk Podcast (2026-08-06). Private / mobile-app names: Pearl Diver (DSP data partner), AppsFlyer, and Vibe (now Walmart) featured in ad-tech discussions but are not separately listed US tickers.