Newsletter · · Ashutosh Agarwal

Agentic Ads Dominate Cannes as Meta Jumps on Its Cloud Pivot - Weekly Online Ads Podcast Recap - Week of June 29 to July 6, 2026

Cannes Lions wrapped with agentic commerce and retail media eating the conversation, from Amazon's Alexa+ agentic ads to Walmart Connect's Vibe deal and Criteo doubling its ChatGPT advertiser base, while Meta jumped 10% on a surprise cloud pivot. Our synthesis of the online-ads podcast tape for the week of June 29 to July 6, 2026.

Weekly Online Ads Podcast Recap

Week of June 29 to July 6, 2026: Agentic Ads Dominate Cannes as Meta Jumps on Its Cloud Pivot

Cannes Lions wrapped this week and the whole industry talked past the creative work to a single idea: the ad you never leave. Agentic commerce and retail media ate the conversation, while Meta's surprise cloud pivot handed the tape its one clean single-stock story.

Executive Summary

  • "Agentic ads" were the defining Cannes narrative. Amazon launched Alexa+ agentic ads ("you see an ad, you ask it a question, you complete the purchase, and you never leave the ad"), collapsing awareness, consideration and conversion into "one moment." A VML exec's line summed up the week: "the recommendation is becoming the new shelf."
  • Retail media is pitched as advertising's fastest-growing channel at ~$150B, but that figure "comes from those who are selling it." Walmart Connect dominated, buying SMB ad platform Vibe to funnel ~10,000 advertisers into CTV, a move said to "point straight at Meta and Pinterest."
  • OpenAI's ChatGPT became a real ad channel, and Criteo (CRTO) is the plumbing. Criteo is now at "just over 2,000" advertisers inside ChatGPT (doubled in ~5 weeks), with "80% of the referred traffic... new to brand."
  • Meta (META) jumped ~10% on 7/1 on its cloud/compute-rental launch. Bulls (Weiss, Terranova) called it the long-awaited proof that Meta can monetize its AI capex, "15 times this year, 13 times next year... compellingly cheap." Skeptics note the cloud business doesn't exist yet and capex is still rising.
  • The Google (GOOGL) ad-tech remedy decision is overdue and expected in 2H. DOJ wants a structural ad-exchange divestiture; the base case handicapped on-air is "about a 60 percent chance that the remedies are only behavioral."
  • New cookie-deprecation research quantified the pain: killing third-party cookies cuts publisher ad prices ~29% to 35%; Privacy Sandbox recovered "only 4%" and doubled ad latency (~3% of impressions fail). EU damage is roughly 2x the US, driven by GDPR.
  • AI ad tooling went from experiment to default across Meta, Google, TikTok (Symphony Agent), Amazon and Canva (Grow 2.0), even as Cannes awards celebrated human, in-camera craft ("AI doesn't buy you good taste") and a Typeface study found campaigns are taking longer to ship.
  • Battleground valuations debated: Adobe (ADBE) "melting ice cube vs. mispriced survivor" (10th straight quarter of ARR deceleration); AppLovin (APP) "reasonable valuation" but needs 50%+ revenue growth to re-rate; Snap (SNAP) after a "billion dollar drop in valuation."

Dominant Themes

1. Agentic commerce and "ads you never leave" were the defining Cannes narrative. Nearly every panel converged on "agentic." Amazon's headline move was launching Alexa+ agentic ads (beta): "you see an ad, you ask it a question, you complete the purchase, and you never leave the ad," with Papa John's and a Sony ticketing arm testing, timed to Prime Day (The Watson Weekly: eCommerce Strategy & News, 2026-06-29). The shared point of view: the funnel is collapsing into "one moment" (The Marketing AI SparkCast with Aby Varma, 2026-06-30). A VML exec: "the recommendation is becoming the new shelf."

2. Retail media is now framed as the fastest-growing ad channel, at ~$150B, but the number comes from the sellers. "The industry puts the number near $150 billion, the fastest growing channel in advertising, and that figure comes from those who are selling it" (The Watson Weekly, 2026-06-29). Walmart Connect dominated: at Cannes with Vizio and a Tan France-hosted shoppable TV, and acquiring Vibe to funnel its ~10,000 marketplace advertisers into CTV, "point[ing] straight at Meta and Pinterest."

3. AI ad-creation tooling is going from "experiment" to "default." Cannes launches: TikTok's Symphony Agent, Amazon Alexa+ agentic ads, Canva Grow 2.0 (create, then publish to TikTok/LinkedIn/Meta, then optimize inside Canva; Canva also acquired Magic Brief), and Google baking AI into smart bidding (The Marketing AI SparkCast, 2026-06-30; Everyday AI Podcast, 2026-07-01). Consensus: "AI is the default and not the experiment... they're becoming the baseline."

4. OpenAI's ChatGPT crossed into being a real ad channel, and Criteo is the plumbing. OpenAI attended Cannes for the first time and "made a statement about they're clearly in the advertising business," citing "one in five queries carry commercial intent." Criteo said it now has "just over 2,000" advertisers running inside ChatGPT, "up from... 1,000 sometime in May... it doubled in about five weeks," with "80% of the referred traffic coming from paid advertising on OpenAI are new to brand" (Marketecture: Get Smart. Fast., 2026-06-29).

5. Measurement is being rebuilt off a "broken cookie spine" toward identity-based, full-funnel, incrementality-first models. Walmart Connect's Job Gibbs argued measurement "has regressed" over 10 years as the identity spine degraded; CTV-exposed audiences drove "an average 28% increase in brand search terms" and "a 30% ad to cart increase" versus pre-exposure (The Commerce Collective Podcast, 2026-06-29). Criteo's Komasinski echoed the anti-"last-click ROAS" line: "most marketers believe that they've been overinvested in last click ROAS for many, many years... harvesting demand that they were already going to get anyway."

6. Counter-melody: AI scale versus human taste, "results not capability." The awards rewarded human craft: Apple TV's in-camera glass-sculpture spot won the Design Grand Prix ("gloriously human made"); Coinbase won Film Craft for a physically-built video-game world. GaryVee's and the NBCU/State Farm/Autodesk/Coinbase panels landed on: "AI doesn't buy you good taste." A Typeface study found campaigns taking longer to ship even with AI (a third of firms needing one to two months).

Active Debates

1. Meta's cloud pivot: proof of capex ROI, or a stock chasing a narrative? Meta announced a cloud business to rent AI compute (for example, to xAI), sending the stock up ~10% on 7/1 (Halftime Report, 2026-07-01).

  • Bull (Steve Weiss, CNBC): "15 times this year, 13 times next year... those are compellingly cheap"; he bought "pretty significantly." The move answers "the number one question... how are they going to generate return on that capex spent... Now they have cloud... That question's been answered."
  • Bull (Joe Terranova, CNBC): compute rental is "a very powerful demand-oriented business," a "paradigm shift" and "the justification for monetizing the spend."
  • Cautious: Weiss's "biggest concern was they announced an increase to their capex this quarter"; Meta "had a horrible first half"; cloud "is not going to happen today."

2. The Google ad-tech remedy: structural divestiture versus behavioral-only. DOJ won an April-2025 liability verdict (unlawful monopolization of the ad-tech stack via tying of publisher ad server and ad exchange); the remedies phase closed Oct-2025, and the decision is overdue (Votes & Verdicts, 2026-07-02).

  • DOJ: seeking "divestiture of Google's ad exchange and maybe parts of its publisher ad server."
  • Analyst base case: "about a 60 percent chance that the remedies are only behavioral," banning tying and requiring non-discriminatory routing. Expected 2H 2026, "any day," though timing has slipped past the judge's January signal.

3. Cookie deprecation: how much does it cost publishers, and can Privacy Sandbox backfill it? New research (Garrett Johnson and Shunto Kobayashi, 5,000+ publishers via Raptive) (Mobile Dev Memo Podcast, 2026-06-30):

  • Cost: removing third-party cookies cuts ad prices ~29% (35% on cookie-carrying impressions); Sandbox recovered "only 4%"; latency doubled (~2s versus ~1s), failing ~3% of impressions. EU damage ~2x US (~66% versus ~33% of cookie-impression revenue), driven by GDPR.
  • "Industry will innovate" counter, rebutted: the effect is already net of adaptation (hashed emails, probabilistic IDs), and historically the industry "innovates towards more identifiers... not contextual advertising," implying a hard privacy-versus-revenue trade-off.

4. Is Adobe a melting ice cube or a mispriced survivor? Roughly a 10th straight quarter of decelerating ARR; the stock hit multi-year lows near $190, then recovered to ~$220 (~10 to 12x) (The Synopsis, 2026-07-03).

  • Bear (Speaker A): "Adobe's kind of reminiscent of a melting ice cube... I can't get over it." Cites gen-AI competition, price increases delayed until after the CEO exits, a CFO departure to Marvell, and a "purposefully confusing" resegmentation.
  • Bull (Speaker B): "AI is creating different images. And Adobe is not an image creator, it's an image modifier." AI editing "sucks as an experience" for precision work; Adobe stays dominant in pro Photoshop/Lightroom, has a $25B buyback and low-double-digit growth, and the "zero or a win" binary is overdone.

5. AppLovin: reasonable value or triple-digit froth? (Halftime Report, 2026-07-01) Terranova: "this is a reasonable valuation... A lot of people think this is triple-digit valuation. It is not." But: "They have to drive the revenue growth back higher, above 50% once again. EPS growth... around 80%. You're not going to get the triple-digit gains that you got in the past."

6. Do LLM/agentic ads create incremental demand or just cannibalize budgets? (Marketecture, 2026-06-29) Criteo's Komasinski: the 80%-new-to-brand stat argues you should "carve off budget from other brand type of budgets," yet he asks himself: "does the growth in the business actually change... or you're moving the bottom of the funnel just a teeny bit?" He is "very staunchly against... the autonomous buying scenario" but bullish on assisted and ambient use cases.

7. Fully autonomous AI buying: imminent or overhyped? The Watson Weekly host warned Cannes was "capability and intent. Very little of it was actual results," flagging: "When an agent sits between your brand and your customer, who decides what the agent recommends?" OpenAI notably "walked away from... owning the checkout."

Tickers Mentioned

META, Meta Platforms. Bull: the cloud/compute-rental launch answers the capex-monetization question; "15 times this year, 13 times next year... compellingly cheap" (Weiss); "paradigm shift... justification for monetizing the spend" (Terranova). Options desk: "the most bullish [trade] is in Meta, now up almost 10%... more than a billion dollars in premium traded... call volume is exploding 20 times the five-day average" (Oliver Rennick). Bear: rising capex; "horrible first half"; cloud "not going to happen today." Source: Halftime Report, 2026-07-01 (Steve Weiss, Joe Terranova, Oliver Rennick, CNBC). Also: Walmart's Vibe deal "point straight at Meta and Pinterest" (The Watson Weekly, 2026-06-29).

GOOGL / GOOG, Alphabet. Bull: Belsky's final trade was "Google"; Google AI ad tooling is embedded as an industry baseline. Bear/overhang: the DOJ ad-tech remedy, with DOJ seeking an ad-exchange divestiture; the analyst gives "about a 60 percent chance that the remedies are only behavioral," decision 2H 2026. Sources: Votes & Verdicts, 2026-07-02; Halftime Report, 2026-07-01 (Belsky, BMO). Quote: "about a 60 percent chance that the remedies are only behavioral... allowing publishers and advertisers to freely use any... rival software all along that chain."

AMZN, Amazon. Bull (ads): launched Alexa+ agentic ads at Cannes; pulled Prime Day into June to "own a stretch of calendar that no one else controls," feeding "the advertising business, which has quietly... become one of Amazon's fattest margins"; Adobe projected $26.3B US online spend. Bear/macro: AOVs "down 17% with more shoppers reaching for buy now pay later"; CPI +4.2% in May. Source: The Watson Weekly, 2026-06-29; The Marketing AI SparkCast, 2026-06-30. Quote: "Amazon launched Alexa Plus Agentic ads. You see an ad, you ask it a question, you complete the purchase, and you never leave the ad."

WMT, Walmart (Walmart Connect). Bull: CTV-exposed audiences drove "an average 28% increase in brand search terms" and "a 30% ad to cart increase"; ~150M customers plus stores where "about 70% of total retail sales... are happening in stores" while "about 83% of ad spend is online"; buying Vibe, plus shoppable TV with Tan France. Bear (the stock): "Walmart technically looks awful... down 4.5%" on soft near-term trends (Terranova, Halftime Report). Sources: Job Gibbs, Walmart Connect, The Commerce Collective Podcast, 2026-06-29; Mike, Walmart Connect, Next in Media, 2026-06-30. Quote: "Advertisers saw an average 28% increase in brand search terms among CTV exposed audiences... And they saw on average a 30% ad to cart increase." (Gibbs)

RDDT, Reddit. Bull: "the validation layer," with "79% of Redditors say they do more research before making a purchase now than they did a year ago" and "62% of Redditors say that a thread debating the pros and cons of something gives them more confidence... more than reviews, more than influencers"; a "Mission Mindset" report with WPP Media; "the number one indexed platform for AI." Bear/friction: "the majority of campaigns have comments off" because "every space on Reddit needs to be moderated... a lot of work"; astroturfing risk. Sources: AdExchanger, 2026-06-30; The Art of the Brand, 2026-07-02. Quote: "79% of Redditors say they do more research before making a purchase now than they did a year ago."

APP, AppLovin. Bull: "this is a reasonable valuation... A lot of people think this is triple-digit valuation. It is not"; a BTIG 2H pick; up 10% that day. Bear/bar: "They have to drive the revenue growth back higher, above 50% once again. EPS growth... somewhere around 80%. You're not going to get the triple-digit gains that you got in the past." Source: Halftime Report, 2026-07-01 (Joe Terranova, CNBC).

SPOT, Spotify. Bull: a B of A Q3 pick; "revenues are still very, very strong"; the product is "better than Apple" for music and podcast. Bear: "caught up... in that whole AI disintermediation conversation"; underperforming. Source: Halftime Report, 2026-07-01 (Brian Belsky, BMO). Quote: "Spotify has been underperforming, but if you take a look at where the cash flows are coming in and revenues are still very, very strong... it's actually better than Apple."

SNAP, Snap. Bull: Snapchat Specs reduce friction ("you just take a picture and say... send it to my friends"); native daily users succeed, implying a distribution runway; the speaker intends to "go take advantage of that billion dollar drop in valuation because I think the product has the legs." Bear: the "billion dollar drop in valuation"; box-checking marketers fail. Sources: The Art of the Brand, 2026-07-02 (Snap CMO Grace Kao also on the GaryVee AI panel, 2026-06-29). Quote: "I'm going to go take advantage of that billion dollar drop in valuation because I think the product has the legs."

ADBE, Adobe. Bull: "Adobe is not an image creator, it's an image modifier"; dominant in pro Photoshop/Lightroom; ~10 to 12x; a $25B buyback; low-double-digit growth; Adobe Brand Studio (VP Rashawn Dhanraj) using AI to free creatives to be "better strategists." Bear: "melting ice cube"; 10th straight quarter of ARR deceleration; price hikes delayed past the CEO exit; the CFO left for Marvell; multi-year lows ($190 to ~$220). Sources: The Synopsis, 2026-07-03; The GaryVee Audio Experience, 2026-06-29. Quotes: "AI is creating different images. And Adobe is not an image creator, it's an image modifier." / "Adobe's kind of reminiscent of a melting ice cube to me, and I can't get over it."

CRTO, Criteo. Bull: repositioning into commerce/discovery plus LLM ads; ChatGPT advertisers "just over 2,000... doubled in about five weeks," "80% of the referred traffic... new to brand"; an Albertsons chatbot-ads deal; buy- and sell-side "a real benefit." Bear: mostly "test budgets today"; incrementality unproven; historically under-invested in UI. Source: Michael Komasinski (Criteo), Marketecture: Get Smart. Fast., 2026-06-29. Quotes: "Just over 2,000, which was up from, like, 1,000 sometime in May. So it doubled in about five weeks." / "80% of the referred traffic coming from paid advertising on OpenAI are new to brand."

PINS, Pinterest. Named in passing as a target of Walmart Connect's SMB push (Vibe's ~10,000 advertisers "point straight at Meta and Pinterest"). Source: The Watson Weekly, 2026-06-29.

Also on the tape (private and adjacent names): OpenAI (ChatGPT ads, "clearly in the advertising business," "one in five queries carry commercial intent"; The Marketing AI SparkCast, 2026-06-30). Canva, with Grow 2.0 and a Magic Brief acquisition (Everyday AI Podcast, 2026-07-01). TikTok/ByteDance, with Symphony Agent. Vizio (Walmart-owned), "30 minutes per day... searching and discovering on our home screen," "the front door in the living room" (Next in Media, 2026-06-30). WPP Media, which co-produced Reddit's "Mission Mindset" report (AdExchanger, 2026-06-30). Dentsu/Carrot, which ran the Papa John's, Instacart and NBCUniversal "empty fridge" campaign (The Watson Weekly, 2026-06-29). Publicis/Digitas, with Amy Lanzi's "the CMO is a dying role" (Decoder with Nilay Patel, 2026-07-02).

BABA, Alibaba (tangential, non-ad). Discussed re: a "$600 million Department of Justice settlement... violations of the Food, Drug, and Cosmetic Act," with concern that platform liability "sets a problematic precedent." Separately, K-Web (China ETF) saw heavy bullish options flow after China PMI "flipped positive." No ad-business commentary. Sources: Simply Trade, 2026-07-03; Halftime Report, 2026-07-01.

Gaps and What Was Not Covered

  • CTV pure-plays: no substantive TTD, ROKU or YouTube CTV commentary (CTV ran through Walmart Connect and Vizio).
  • Streaming ad tiers: no NFLX or DIS ad-tier subscriber or ARPU discussion.
  • Programmatic and verification: no MGNI, PUBM, DV or IAS.
  • Agency holdcos: the OMC-IPG merger, WPP, PUB FP and S4 got only passing mentions.
  • Apple Search Ads and App Store ads (AAPL): no coverage.
  • Ad-spend macro forecasts: no GroupM, Magna, Zenith or eMarketer figures.
  • Chinese ad platforms: BIDU, PDD, TCEHY, NTES and JD saw no substantive ad-business discussion; BABA appeared only in a non-ad regulatory context.