Newsletter · · Ashutosh Agarwal

AI Overviews and Alphabet Capex Rattle Google's Ad Dominance - Weekly Online Ads Podcast Recap - Week of July 27, 2026

A recap of the online advertising podcast conversation for the week of July 27, 2026, centered on Alphabet's earnings and the 'Google Zero' AI Overviews threat, retail media's shift to proving incrementality, and connected TV crossing into a performance channel.

Weekly Online Ads Podcast Recap

Week of July 27, 2026: AI Overviews and Alphabet Capex Rattle Google's Ad Dominance

The conversation was unusually concentrated: this was the week Alphabet reported earnings and the market's decade-old assumption, that Google's ad machine is untouchable and "asset-light", cracked in public. Around that, three big storylines ran in parallel: retail media "growing up" and being forced to prove it actually works, connected TV crossing over from a brand-awareness channel into a performance channel, and the slow-motion collapse of the deal between Google and the websites it has always depended on ("Google Zero").

Executive Summary

  • "Google Zero" went mainstream. After Alphabet's earnings, podcasters across finance and tech shows converged on one worry: Google's AI Overviews answer the question on the results page, so people stop clicking through to websites. A new academic field experiment discussed on the Mobile Dev Memo Podcast (July 21) put a hard number on it, showing an AI Overview cuts click-throughs to other websites by about 40%, with no measurable improvement in the user's experience. On The Vergecast (July 24), the hosts ran through a growing list of publishers (USA Today, Politico, The Economist, Reuters, People Inc.) refusing to feed Google, and even Reddit reconsidering.

  • Reddit vs. Google was the single most-cited news item. Reddit is reportedly weighing whether to renew its ~$60 million-a-year AI-content licensing deal with Google. Reddit was named the #1 cited source for Google's Gemini in July 2026 (Chit Chat Stocks, July 24). Reddit shares fell roughly 8–11% on the report (Brew Markets, July 22). Most commentators read the leak as a negotiating tactic, and think Reddit holds the stronger hand.

  • Alphabet's earnings reframed it from an ad company to a capital-spending company. Cloud revenue up 82%, total revenue up 24%, backlog up to $514 billion, but the stock fell ~7% because Alphabet raised its AI capital-spending guidance to over $200 billion and, per The Canadian Investor (July 25), posted its first free-cash-flow-negative quarter since 2015 and would not answer analysts' repeated question about the return on that spending.

  • Retail media is "off the family phone plan." Criteo's head of retail media pegged the market at roughly $75 billion, with Amazon holding about two-thirds of it, growing 18–20% a year versus 2–3% for traditional retail (The CPG View, July 21). The theme everywhere: brands have stopped asking "does retail media work?" and now demand proof of incrementality, real, causal sales lift.

  • Connected TV is becoming a performance channel. 70% of media buyers plan to raise CTV budgets next year (MarTech Podcast, July 20). Disney said 95% of its streaming ad campaigns are now sold against a specific audience rather than a show, and more than half of its $5 billion streaming ad business is now transacted programmatically (The Big Impression, July 22).

  • AppLovin (APP) "works", grudging verdict from a skeptic. Independent measurement firm Haus, long a doubter, concluded AppLovin genuinely drives sales, because its ads produce fast results and it made independent incrementality testing easy (Marketecture, July 24). It works best for low-price, impulse products like apparel and footwear.

  • Agencies are rebranding to survive AI. Omnicom's North America media CEO argued the term "holding company" is outdated, they are now "capability companies" selling strategy and first-party-data expertise, not just cheap bulk ad-buying (The Digiday Podcast, July 21), pointing to the IPG acquisition as evidence.

  • Everyone's endgame is the transaction. From Google's AI-mode-to-Instacart-cart integration to Stripe's reported $53 billion bid for PayPal, the week's throughline was platforms racing to own the moment of purchase, because if AI answers the question and an agent completes the buy, whoever owns the checkout owns the money (Selling on Giants, July 21).

  • Most-discussed tickers: GOOGL/GOOG, META, AMZN (including Instacart/CART), RDDT, DIS, APP, SPOT, NFLX, TTD, ROKU, CRTO, OMC, IPG.

Dominant Themes

1. "Google Zero": the search-to-website deal is breaking

The loudest, most-repeated narrative of the week was that Google's own AI features are eating the traffic that used to flow from Google to the rest of the internet, and that this now threatens Google's core advertising business, not just publishers'.

The clearest evidence came from an academic study discussed on the Mobile Dev Memo Podcast (July 21), where researchers Ananya Sen and Saharsh Agarwal described a real field experiment run through a custom browser extension on about 1,000 users. Rather than just comparing before-and-after (which they dismissed as merely "correlational"), they randomly hid or showed Google's AI Overviews so they could measure true cause and effect. The result: "showing AI overviews reduces clicks by 40% conditional on the search happening, which is a fairly big number." Crucially, they pushed back on Google's defense that it only removes low-value clicks: "we find that the quality of the clicks were actually not different across the two groups." And the twist, users didn't notice or mind: "surprisingly, user experience was absolutely unchanged, even when we hid the AI overviews… users in both groups felt exactly the same." In other words, AI Overviews are great for Google and its users, and quietly devastating for everyone downstream. Mobile Dev Memo Podcast

On The Vergecast (July 24), the hosts turned this into a full "you can't ignore Google Zero anymore" segment. They noted that giving content to Google is becoming "economically irrational and everyone will stop doing it," and rattled off publishers pulling back or suing, "USA Today, Politico, The Economist, People Inc., Reuters", with Penske Media "openly suing Google over AI Overviews not delivering any traffic." The deeper point: Google has tied AI training to search indexing, so "if you want to not be crawled by Google's AI systems, you can't be crawled by its search systems." They cited a Cloudflare report that more than half of internet traffic is now non-human bots, and that "for every hour spent online searching for information, only 15 minutes is spent on the open web." The most quotable line captured the paradox facing Google: "Google is creating a world in which its own best business model doesn't work anymore… Google will tell you that people click the links, and I can tell you from my own experience and from all of these publishers and from Reddit, people aren't clicking the links." The Vergecast

2. Alphabet's earnings: the market suddenly sees a capital-heavy company

Alphabet reported during the week, and the reaction was the dominant investing story. On The Canadian Investor (July 25), the hosts laid out why a company posting "cloud revenue up 82%, revenue up 24%, backlog up 10% to $514 billion" could still drop about 7%, hundreds of billions in market value: "it really comes down to CapEx guidance." Since first guiding 2026 capital spending, Alphabet has "increased it by 11%" in six months, to over $200 billion, and this was "the first quarter that they were free cash flow negative" going back to 2015. The most damning detail: management was "asked a few times on the call, what's the return on invested capital for this AI spend? And they did not answer." The framing that stuck: Google is "changing from a business that for the last decade or so… was an asset-light business, probably the best ad business that the world has ever seen" into one that is "issuing equity… issuing debt," with Meta "in the same boat." The Canadian Investor

The Vergecast tied the earnings back to the ad worry: Alphabet actually missed slightly on search, "they missed by, like a hundredth of a billion dollars", but for a business that "has been immune to everything," even "any teeny tiny hole in that case starts to look like a big scary problem." The Vergecast

3. Retail media "grows up" and is forced to prove itself

The retail media network (RMN), the ads that big retailers sell against their own shopper data, was the second-biggest theme, framed everywhere as an industry maturing from land-grab to accountability. On The CPG View (July 21), Criteo's president of retail media Sherry Smith sized it plainly: the market is "about 75 billion or so," with "about two thirds of that… Amazon," growing "18 to 20% CAGR compared to 2 to 3% for traditional retail and 8% for e-commerce." Her core message: "retail media is absolutely moving from expansion to accountability… The question is not just, can you get into the game, but… can you measure incrementality? Can you prove consistent value to brands?" Her memorable line for how brands now think: "brands are no longer asking whether retail media works. They're asking how well it works. And they're asking compared to what?" She warned that "launching an inventory alone does not create a media business," and named the mature players, "Amazon, Walmart, even like a Best Buy and a Target, they've been in this business for over a decade." The CPG View

Instacart made the same "prove it" case from the platform side. On The CPG Guys (July 22, live from Cannes Lions), Ali Miller and Tim Castelli leaned on Instacart's "receipt-level purchase verified deterministic truth", because everything is digital, they see the actual basket, substitutions and searches. Instacart was "one of the first retail media networks that was accredited by the MRC" (a measurement watchdog) and worked with Circana to prove impact. The strategic shift: Instacart is taking its ad technology off-platform, letting brands use its first-party data to target and measure across "TikTok, YouTube, Meta, The Trade Desk, Roku, NBCUniversal, so many other places," reaching "the 2,200 retailers on our marketplace and the over 310 partners that are using our CaratOut solutions." Notably, they said they are intentionally not yet putting ads inside their AI shopping experiences. The CPG Guys

4. Connected TV becomes a performance channel

The idea that TV advertising can now be measured and optimized like digital, not just used for brand awareness, ran through several shows. On the MarTech Podcast (July 20), AdRoll's Allison Clark opened with the stat of the week: "70% of media buyers expect to increase their connected TV budgets next year," driven by "an AI-driven playbook where there's less site traffic and weakening digital direct response attribution." Her distinction: "linear is still very much an awareness play," but CTV "is crossing over into that performance marketing, where you can measure and see those business outcomes, see conversions." She also flagged the same click-death theme, "There is no clicks in CTV", meaning advertisers must retarget from ad exposure rather than a click. MarTech Podcast

Disney provided the marquee CTV data point. On The Big Impression (July 22), Jamie Power, Disney's SVP of Addressable Sales, explained the shift from buying shows to buying audiences: "95% of campaigns that are brought in streaming have an audience assigned to it," enabled by unifying the ad server across Disney+, Hulu and ESPN. On money: "way more than half of our $5 billion streaming business is transacted programmatically," and "biddable actually is the fastest growing way that clients are choosing to transact their upfront commitments." Her "Compass" data-collaboration system lets Disney send first-party data into a DSP "like the trade desk" for measurement. The Big Impression

On Next in Media (July 23), Caden's Doug Rosen argued the industry is over-indexing on connection and under-indexing on context: "there's too much emphasis on the notion of connections and not enough on the contextualization of the video… that contextual is what will drive the performance more than just the connection." He called YouTube "the ultimate destination for contextual television" and pushed "total video" planning, "one plan, one audience, one time", to fix the universal complaint that the same ad follows viewers everywhere: "we're not sitting there saying, oh my God, that brand has great frequency… We sit there and say, make it stop." On the ad market: the second half feels "opportunistic," helped by the World Cup, the NBA playoffs, and political spending. Next in Media

5. AI is redrawing who wins in advertising, and everyone's endgame is the transaction

On Motley Fool Hidden Gems Investing (July 20), the panel framed the structural picture: "Alphabet and Meta control about half of the entire global advertising market. For Alphabet, advertising accounts for about 70% of its total business… Meta relies on advertising for about 98% of its entire revenue," combining into roughly a "$6 trillion market cap." Their counterintuitive point was that AI, feared as a threat, has instead "triggered an advertising boom" by rewarding whoever controls consumer data and can afford to build automated ad infrastructure. Motley Fool Hidden Gems Investing

Selling on Giants (July 21) tied the transaction theme together: Google's new AI-mode integration lets a shopper "plan something like a barbecue, build the grocery list, and add the ingredients into an Instacart cart", "Search is becoming an action layer… Machine-readable product data is becoming shelf placement." And Stripe's reported "more than $53 billion" bid for PayPal matters because "the platform that owns the wallet can influence commerce just as much as the platform that owns discovery." Selling on Giants

Active Debates

Debate 1, Is Alphabet a screaming buy on the dip, or a value trap?

The bull view (Chit Chat Stocks, July 24) is straightforward: cloud is accelerating, backlog is enormous, and the sell-off is an over-reaction, "buy the dip on Netflix and Google." The bear view (The Canadian Investor, July 25) is that investors "jumping on the opportunity to buy it are just overlooking the potential risk" that $200 billion+ of annual AI spending simply may not earn its cost of capital, and management would not even answer the return-on-investment question. One host went further, arguing Google is oddly behind on model quality despite owning the most data: "I don't understand how Google is so far behind on the development of actual models." Both sides agree on the facts (huge cloud growth, huge capex); they disagree on whether the spending is a moat or a money pit. Chit Chat Stocks · The Canadian Investor

Debate 2, Does AI kill the search-ad model, or just move it?

On Motley Fool Hidden Gems (July 20), one analyst argued the duopoly is insulated: "the alphabets and metas of the world remain the dominant winners because their ecosystems and platforms are very well insulated against AI disruption." A co-host pushed back on the duopoly framing: "don't count out Amazon… its ads sit a lot closer to the actual purchases you make than either Google or Facebook ads do," and flagged "the real wild card here is agentic AI when it comes to shopping… The question of who gets paid when an AI agent clicks buy, there's not really a clear answer there just yet." The Vergecast made the darker version of the same argument, that AI Overviews may hollow out the very content ecosystem Google needs. Motley Fool Hidden Gems Investing

Debate 3, Reddit vs. Google: who needs whom more?

Everyone agreed Reddit is publicly floating that it might not renew its ~$60M/year Google deal, but they split on interpretation. Brew Markets (July 22) read it as leverage: "the public airing of its concerns frankly feels like a negotiating tactic aimed at securing a richer deal from Google," noting $60M "mostly drops straight to the bottom line" against Reddit's "$520 million plus of net income last year" and "roughly 1.5 billion monthly average users." Chit Chat Stocks (July 24) argued Reddit holds the cards because it's "the #1 cited source" for Gemini and "OpenAI is already paying more than that," concluding "you could probably charge each of these companies hundreds of millions of dollars a year." The skeptical counter-question from a listener: "I still don't understand why we decided Reddit's data was the holy grail", with a nod to astroturfing and fake content. Brew Markets · Chit Chat Stocks

Debate 4, Does AppLovin actually work, and for whom?

Marketecture (July 24) tackled this head-on with a self-described former skeptic, Haus's Olivia Kory. Verdict: "It does" work, but the nuance is the debate. Her theory is that AppLovin's edge is speed: "the effects of AppLovin Ads are short-term. They work immediately," which suits performance marketers who "want to see immediate ad effects." The corollary is that it's not universal, it shines for "lower AOV, more impulse purchases like apparel, footwear," while "subscription products are a little bit harder because that's a longer-term commitment." She credited AppLovin for making independent verification easy from the start: "setting up an incrementality test with AppLovin is as easy as it is in any like Meta or Google." Marketecture

Debate 5, Can agency holding companies survive AI, or do consultancies eat them?

On The Digiday Podcast (July 21), Omnicom Media North America CEO Ralph Pardo rejected the "holdco" label entirely: "We're not holding anything. We are actually capability companies." The bull case is that when AI does the grunt work, clients still need high-level strategists and first-party-data gatekeepers. The bear case, voiced by the Digiday hosts: if AI can do the media-buying, "why am I paying you a shit ton of money to do this?", and the specialization pivot walks agencies straight into competition with consultancies like "Accenture Song… McKinsey, Boston Consulting." One host's blunt summary of the current strategy: "we're in a wet spaghetti noodle at the wall kind of stage as they figure this out amidst things like layoffs." The Digiday Podcast

Debate 6, Contextual vs. audience targeting in CTV

Next in Media (July 23) surfaced a genuine philosophical split in TV buying: is precision (know exactly who the viewer is) or context (know what they're watching) the bigger driver of performance? Doug Rosen argued the pendulum has swung too far toward connection/precision and that "what's old is new", contextual targeting, which worked in early digital, drives outcomes and solves the ad-frequency problem. This runs against the prevailing Disney-style "buy the audience, not the show" model, suggesting the two approaches will be layered rather than one winning outright. Next in Media

Tickers Mentioned

GOOGL / GOOG, Alphabet

  • Bull: Cloud revenue up 82%, total revenue up 24%, backlog up to $514B; the post-earnings dip is a buying opportunity as cloud accelerates alongside AWS and Microsoft. AI-driven ad targeting has helped, not hurt, the ad business.

  • Bear: AI Overviews cut outbound clicks ~40% and are hollowing out the web Google relies on; capex guidance raised to $200B+, first FCF-negative quarter since 2015, and management won't quantify the return on AI spend; a slight search-revenue miss dents the "immune to everything" story.

  • Source episodes & speakers: "You can't ignore Google Zero anymore," The Vergecast (July 24), Nilay Patel / David Pierce / Alex Cranz, The Verge; " Higher Yields, Canada's Housing Gridlock, and Investors Turn on Google," The Canadian Investor (July 25); "Quantifying the impact of AI Overviews on outbound clicks," Mobile Dev Memo Podcast (July 21), Ananya Sen & Saharsh Agarwal (academic researchers); "Buy The Dip On Netflix And Google?," Chit Chat Stocks (July 24); "Chinese AI Just Made History," Motley Fool Hidden Gems Investing (July 20).

  • Direct quotes: "Google is creating a world in which its own best business model doesn't work anymore… people aren't clicking the links. They're not." (The Vergecast) · "showing AI overviews reduces clicks by 40% conditional on the search happening." (Mobile Dev Memo) · "it really comes down to CapEx guidance… the first quarter that they were free cash flow negative… what's the return on invested capital for this AI spend? And they did not answer." (The Canadian Investor)

  • Links: The Vergecast · The Canadian Investor · Mobile Dev Memo Podcast · Chit Chat Stocks · Motley Fool Hidden Gems Investing

META, Meta Platforms

  • Bull: Roughly half the global ad market shared with Alphabet; ~98% of revenue from advertising; AI-driven automated ad tools and unmatched consumer data make it a primary beneficiary of the AI ad boom.

  • Bear: Now "in the same boat" as Google, moving from an asset-light ad machine toward heavy AI capital spending and debt/equity issuance, with returns unproven; one investor's benchmarking found Meta's models (Muse/Spark) underperforming.

  • Source episodes & speakers: "Chinese AI Just Made History," Motley Fool Hidden Gems Investing (July 20), Rachel & Matt (Motley Fool analysts); "Investors Turn on Google," The Canadian Investor (July 25).

  • Direct quotes: "Meta relies on advertising for about 98% of its entire revenue." (Motley Fool) · "Meta's in the same boat, right? So you had these businesses that were asset light… and now you're looking at a business that's spending more and more money." (The Canadian Investor)

  • Links: Motley Fool Hidden Gems Investing · The Canadian Investor

AMZN, Amazon (Amazon Ads, plus Instacart/CART ecosystem)

  • Bull: Amazon's ad business "is rapidly becoming really a third major player in ads," with ads sitting closer to the actual purchase than Google or Meta; commands ~two-thirds of the ~$75B retail media market; AWS cloud likely accelerating.

  • Bear: Heavy capital investment will pressure the stock near-term; agentic shopping could scramble who gets credit (and paid) for a sale.

  • Source episodes & speakers: "Chinese AI Just Made History," Motley Fool Hidden Gems Investing (July 20); "Winning the Next Chapter of Retail Media Networks," The CPG View (July 21), Sherry Smith (President, Retail Media, Criteo); "Buy The Dip On Netflix And Google?," Chit Chat Stocks (July 24).

  • Direct quotes: "don't count out Amazon. Amazon's ad business… it's rapidly becoming really a third major player in ads and its ads sit a lot closer to the actual purchases you make than either Google or Facebook ads do." (Motley Fool) · "you've got about two thirds of that's Amazon." (The CPG View)

  • Links: Motley Fool Hidden Gems Investing · The CPG View · Chit Chat Stocks

RDDT, Reddit

  • Bull: The #1 cited source for Google's Gemini; possesses uniquely valuable human conversation data; ~1.5B monthly average users; strong leverage to raise the ~$60M Google deal (OpenAI reportedly already pays more); can potentially charge multiple AI firms hundreds of millions a year.

  • Bear: The $60M is a small slice of >$2B revenue; walking away risks upsetting users and losing near-pure-margin licensing income; data-quality concerns (astroturfing, fake content); stock fell ~8–11% on the report.

  • Source episodes & speakers: "Reddit Rethinks Google Deal & 3M Beyond the Post-it Note," Brew Markets (July 22), Ann Berry (host); "Buy The Dip On Netflix And Google?," Chit Chat Stocks (July 24); "You can't ignore Google Zero anymore," The Vergecast (July 24).

  • Direct quotes: "Reddit may not renew its AI licensing agreement with Google… that agreement, signed in 2024, is reportedly worth about $60 million a year… Reddit stock dropping by 20 bucks. That's about 11% on the news." (Brew Markets) · "The top 10 cited sources for July 2026 for Gemini: number 1, Reddit… you could probably charge each of these companies hundreds of millions of dollars a year." (Chit Chat Stocks)

  • Links: Brew Markets · Chit Chat Stocks · The Vergecast

DIS, Disney (streaming ad business)

  • Bull: ~$5B streaming ad business with more than half transacted programmatically; a purpose-built, unified ad stack across Disney+/Hulu/ESPN; 95% of streaming campaigns sold against an audience; "biddable" is the fastest-growing upfront transaction method; live sports (record NBA Finals ratings) drive premium, high-attention inventory.

  • Bear: Not framed bearishly this week; the implicit risk is heavy reliance on the same programmatic/DSP plumbing (e.g., The Trade Desk) that commoditizes inventory.

  • Source episode & speaker: "Disney's Jamie Power explains why advertisers are buying audiences, not shows," The Big Impression (July 22), Jamie Power (SVP, Addressable Sales, Disney).

  • Direct quotes: "95% of campaigns that are brought in streaming have an audience assigned to it." · "way more than half of our $5 billion streaming business is transacted programmatically." · "Biddable actually is like the fastest growing way that clients are choosing to transact their upfront commitments."

  • Link: The Big Impression

APP, AppLovin

  • Bull: Independent measurement firm Haus, a former skeptic, confirms it "works," delivering fast, immediately measurable results; made third-party incrementality testing easy from launch; strong fit for low-price impulse e-commerce (apparel, footwear).

  • Bear: Effectiveness is uneven, weaker for higher-consideration or subscription products; the "short-term effects" strength invites the question of whether it's capturing demand that would have converted anyway.

  • Source episode & speaker: "Ask Olivia Kory Whether AppLovin Ads Work & She Explains Attribution vs Incrementality," Marketecture (July 24), Olivia Kory (Haus).

  • Direct quotes: "the effects of AppLovin Ads are short-term. They work immediately." · "setting up an incrementality test with AppLovin is as easy as it is in any like Meta or Google." · "lower AOV, more impulse purchases like apparel, footwear are doing really well in app… subscription products are a little bit harder."

  • Link: Marketecture

SPOT, Spotify

  • Bull: Spotify Ad Exchange ("SAX"), ~18 months old, is meaningfully expanding the pool of active advertisers in audio; built-in generative-AI creative tool lowers the barrier for new advertisers; billions of daily "taste data" signals enable precise targeting; developing first-party brand-lift and incrementality products and better audio attribution (Bayesian MMM).

  • Bear: Not discussed as a bear case; the underlying challenge raised is that audio still lacks a settled attribution model, advertisers struggle to value it correctly.

  • Source episode & speakers: "Inside Spotify's Ad Exchange Takeover," Next in Media (July 21), Per Sandal (Spotify) with host Mike Shields.

  • Direct quote: "we've called that product SAX… the Spotify Ad Exchange has been… incredible for the expansion of active advertisers on the platform… we've got a generative AI creative tool in there if you need help making creative."

  • Link: Next in Media

NFLX, Netflix

  • Bull: Lumped with Google as a post-earnings "buy the dip" candidate; ad-supported tier still developing.

  • Bear: Discussed mainly as a Q2 sell-off; commentators debated whether the pullback is a buying opportunity rather than making a strong fundamental ad case.

  • Source episode & speakers: "Buy The Dip On Netflix And Google? Intel's Rebound Report; Earnings Season Starting Gun," Chit Chat Stocks (July 24), hosts.

  • Direct quote: "buy the dip on Netflix and Google" (episode framing / listener-question discussion).

  • Link: Chit Chat Stocks

TTD, The Trade Desk

  • Bull: Repeatedly cited as the default independent DSP that premium sellers plug into, Disney sends first-party data "into a DSP like the trade desk" via Compass, and Instacart activates its off-platform data through The Trade Desk (among Meta, YouTube, TikTok, Roku, NBCUniversal). Positioned as core plumbing for audience-based CTV and retail-media activation.

  • Bear: Not argued directly this week; the implicit risk is that as platforms (Disney, Instacart, Spotify) build their own exchanges and first-party stacks, independent DSPs must keep proving unique value.

  • Source episodes & speakers: "Disney's Jamie Power…," The Big Impression (July 22), Jamie Power (Disney); "Instacart's Ali Miller and Tim Castelli, Live from Cannes Lions 2026," The CPG Guys (July 22).

  • Direct quotes: "if an advertiser wants to measure a campaign in a DSP in the trade desk, we're able to send our first-party data there." (The Big Impression) · "That includes TikTok… It includes YouTube, Meta, The Trade Desk, Roku, NBCUniversal, so many other places." (The CPG Guys)

  • Links: The Big Impression · The CPG Guys

ROKU, Roku

  • Bull: Named as a key off-platform destination brands activate through Instacart's first-party data and retail-media stack, alongside The Trade Desk, YouTube, Meta and NBCUniversal, i.e., a demand endpoint for retail-media-fueled CTV buying.

  • Bear: No dedicated bear case surfaced; only mentioned in passing as part of the CTV activation ecosystem.

  • Source episode & speakers: "Instacart's Ali Miller and Tim Castelli, Live from Cannes Lions 2026," The CPG Guys (July 22).

  • Direct quote: "That includes TikTok… It includes YouTube, Meta, The Trade Desk, Roku, NBCUniversal, so many other places."

  • Link: The CPG Guys

CRTO, Criteo

  • Bull: Positioned as the neutral arms dealer of retail media, providing the technology retailers use to build and scale networks without the ~decade of investment the leaders made. Benefits from the shift to accountability/incrementality, where independent measurement across networks is prized.

  • Bear: Implicit, the market is heavily concentrated in Amazon (~two-thirds), squeezing everyone competing for the remaining third.

  • Source episode & speaker: "Winning the Next Chapter of Retail Media Networks: Relevance, Loyalty & Execution," The CPG View (July 21), Sherry Smith (President, Retail Media, Criteo).

  • Direct quote: "retail media is absolutely moving from expansion to more and more accountability… brands are no longer asking whether retail media works. They're asking how well it works."

  • Link: The CPG View

OMC, Omnicom (and IPG, Interpublic)

  • Bull: Reframing from "holding company" to "capability company"; the IPG acquisition (plus earlier Flywheel deal) builds scale and first-party-data/strategy expertise; AI handles busywork while agencies move up to higher-value strategic counsel and data gatekeeping.

  • Bear: Clients increasingly ask why they pay agencies at all if AI can do media-buying; the pivot to specialization invites direct competition from consultancies (Accenture Song, McKinsey, BCG); pressure shows in layoffs and "do more with less."

  • Source episode & speakers: "What happens to agencies when AI does the busywork?," The Digiday Podcast (July 21), Ralph Pardo (CEO, Omnicom Media North America) with Kamiko McCoy (Digiday).

  • Direct quotes: "We're not holding anything. We are actually capability companies." · "the whole industry is aware of the IPG acquisition. But prior to that, Flywheel as well." · (host) "if AI can do it, why am I paying you a shit ton of money to do this?"

  • Link: The Digiday Podcast

Other names mentioned in an ad context

  • NBCUniversal (Comcast/CMCSA): cited as an off-platform CTV activation destination for Instacart's retail-media data. The CPG Guys

  • Walmart (WMT), Target (TGT), Best Buy (BBY): named as mature retail media operators "in this business for over a decade," and Walmart/Target as retailers "quietly growing their own sponsored ad revenue." The CPG View · Motley Fool Hidden Gems Investing

  • Outfront Media (OUT): billboard/out-of-home advertising cited as a counterpoint to digital, "the one type of advertising you can't click away." Motley Fool Hidden Gems Investing

  • ByteDance/TikTok (private): TikTok Shop reportedly testing a managed-services pilot ($10k flat fee + 10–20% commission, GMV Max advertising) and flooded with AI-generated shoppable videos; Shark Ninja reportedly banned AI-generated affiliate content in its program. Selling on Giants

  • Innovid (measurement/verification): named as AdRoll's third-party CTV attribution partner. MarTech Podcast