Newsletter · · Ashutosh Agarwal
Digital Ads & Retail Media - Week of June 7, 2026: Meta Set to Pass Google in Ads by Year-End
Digital ads and retail media newsletter for the week of June 7, 2026. eMarketer's call that Meta's ad business overtakes Google's by year-end gets an airing on the podcast tape, alongside the clearest operator evidence yet that agentic ad buying has crossed from pilot to production at brand scale.
Digital Ads & Retail Media
Week of June 7, 2026: Meta Set to Pass Google in Ads by Year-End
TL;DR
- The week's one big number: eMarketer's call, surfaced on FT Tech Tonic, that Meta's ad business overtakes Google's by the end of 2026, and the podcast tape is full of operator evidence for why.
- Agentic buying went from slideware to live spend. A $1B DTC brand now runs 100% of its Meta budget through an AI media buyer, and both Meta and Google are quietly retiring manual ad products. The agency middle is being squeezed in real time.
- Meta-heavy week. Coverage concentrated on Meta and AI-search structure. The podcasts this week did not take up Amazon Ads, The Trade Desk, retail media networks, CTV names, AppLovin, or the ad-measurement complex, so this issue stays on the names that actually drew discussion.
What's new
1. eMarketer's "Meta passes Google" call gets an airing, and a mechanism. On the FT's Tech Tonic - "AI Labs: Zuckerberg's $100bn gamble" (June 3), FT journalist Hannah Murphy noted Meta's ads business "remains strong and in fact is forecast to take over Google by the end of the year by eMarketer," and is bankrolling "hundreds of billions of dollars" of AI spend. Colleague Cristina Criddle added the sharper point: the AI that already runs Meta's ad engine is mature and underappreciated, "it's just not the sexy kind of AI." Why it matters: the bull case on META has quietly shifted from "feed personalization" to "Meta owns the first-party data that OpenAI or Anthropic would love to have."
2. A $1B brand is now 100% agentic on Meta. On the DTC Podcast (June 3), True Classic CEO Ben Diamond disclosed that all of the brand's Meta spend now runs through an agentic AI buyer, live ~6 weeks and "at parity" with human management. "Meta is our largest channel… if we trust this system with our largest channel, it's exciting." He also claims AI creative has "cut our production costs by millions of dollars." This is the clearest operator data point yet that Advantage+-style automation is taking over the actual buying decision, not just assisting it.
3. Google is killing its own manual ad products. On The Marketing AI SparkCast (June 1), Aby Varma reported Google is retiring Dynamic Search Ads in favor of "AI Max," with migration starting "around September" 2026, after already retiring standard text ads and responsive image ads earlier this year. Same episode: Varma says OpenAI launched a self-serve ChatGPT ad manager at ads.openai.com in May ("any brand can buy ads directly inside ChatGPT, cost per click, no minimum spend"), with "WPP, Publicis, Omnicom, Dentsu already integrated." (These are Varma's cited figures, not independently verified, treat as directional.)
4. Meta opens its ad stack to outside AI. Varma also flagged that Meta launched "AI connectors" in beta, letting advertisers "manage Meta campaigns, audiences, creative, and reporting directly through AI tools like Claude and ChatGPT using Meta's own MCP server." On Creator's MBA (June 2), an operator demoed exactly this, pulling live campaign metrics into Claude and having the AI flag that video creative cost "62% more per lead." Small workflow, big read-through: AI assistants may quietly disintermediate attribution vendors.
5. The "impressions and price both up" claim. On Chit Chat Stocks (June 5), Ryan Henderson claimed Meta grew "impressions like high teens and price per ad… high teens percentage" last quarter, "pretty astounding," since the two usually move inversely. He went further: "I would not be surprised if Meta was generating higher advertising revenue than Google… in three, five years." Co-host Brett Schaefer pushed back that Alphabet enjoys the same AI-efficiency tailwind, and called the two "very, very close" at 18x vs. 32x EV/EBIT. (Henderson's growth figures are his recollection of a prior quarter, unverified, flag accordingly.)
The line of the week, from FT's Criddle: if "we're sending off our agent to go do something for us, it will show that agent an ad." Agentic shopping isn't just a threat to the ad model, it may be the next ad surface.
The debate
Bull, AI tooling is expanding the pie and deepening the moat. The walled gardens get better as buying automates: ROAS rises, more budget flows to whoever owns the data and the algorithm, and the brands themselves volunteer that machines now beat humans (True Classic at 100% agentic). Meta's claimed simultaneous lift in both impressions and price-per-ad, if real, is the dream scenario, because it means AI is creating demand, not just reallocating it. First-party, cross-app data is the scarce asset, and Meta and Google sit on the largest pools.
Bear, automation commoditizes the buyer and AI-search hollows the funnel. The same tooling that lifts ROAS strips agencies and attribution vendors of their reason to exist (note the Hyros-displacement aside on Creator's MBA, and SparkCast's warning that AI now handles "the buying and the bidding and the optimization entirely on its own"). More dangerous is the top of the funnel: Varma's cited stats, "almost 60% of all US searches now end without a single click," #1 organic CTR falling "from 27% down to as low as 11%" where an AI answer appears, describe a world where the search ad unit itself is being rebuilt mid-flight. Google retiring its own manual formats is both confident and risky. And OpenAI selling ads inside ChatGPT means a new auction is opening that didn't exist a year ago.
Stocks in play
Meta (META). Bull: AI ad engine is mature and self-funding the capex; operators report automation at parity-or-better; eMarketer sees it passing Google this year. Bear: "98% of revenue is through advertising" (Morning Brew Daily, June 4) is concentration risk, and its non-ad track record is poor, hosts noted ~$80B of metaverse spend "effectively gone," Portal and crypto flops. Watch: Q2 print, whether the impressions-and-pricing-both-up pattern actually shows up in reported family-of-apps revenue and ad pricing.
Alphabet (GOOGL). Bull: same AI-efficiency tailwind as Meta; Berkshire reportedly bought the dip (per Chit Chat Stocks); cheaper at ~18x vs Meta's ~32x EV/EBIT on the hosts' framing. Bear: it is the most exposed to AI-search click destruction and is voluntarily ripping out legacy ad products into "AI Max." Watch: the September AI Max / Dynamic Search Ads migration, and any disclosure on AI Overviews monetization vs. cannibalization.
No other ticker drew substantive podcast discussion this week.
Read-throughs
Bluntly: there weren't any this week. Emerging platforms (RDDT, PINS, SNAP, APP): zero coverage. Retail-media operators (WMT Connect, CART) and the ~$129B retail-media theme: zero coverage, not a single mention. CTV / streaming (Roku, Disney, Netflix, WBD) and the open-internet / TTD Kokai / UID2 story: zero coverage. Ad-measurement vendors (DV, IAS, RAMP, CRTO, MGNI, PUBM): zero coverage. The only adjacent signal worth banking is indirect: privacy-driven measurement pain remains real, Triple Whale's Maxx Blank said Apple's ATT "took the wheels out of advertising in terms of what you could see," which is the structural tailwind under the entire identity/measurement complex even though none of the public names were named.
What changed vs. last week
The center of gravity moved to the operator and practitioner level this week, DTC brands, agencies, and small-business workflows, rather than market-structure or sell-side commentary. The genuinely new, investable thread vs. recent weeks: agentic buying crossed from pilot to production at brand scale (True Classic at 100% of Meta spend), and OpenAI's self-serve ad manager moved from rumor to launched product. The discussion clustered on Meta and AI-search structure, with the names above drawing the substantive coverage.