Refacto

Podcast episode

Episode 178: Cannes Fashion Choices, Plus Patrick Dolan on the State of OOH Advertising

by Ken Rona · ·Marketecture

ai-in-adtech ctv identity m-and-a programmatic

TL;DR

A Cannes-prep episode of Marketecture combining pre-show banter with a substantive interview on out-of-home (OOH) advertising's digital transformation, plus a rapid-fire news segment covering the Fox/Roku deal, Hightouch's bid for LiveRamp's identity assets, and the wave of pre-Cannes AI announcements. Most relevant for publishers and buyers tracking OOH programmatic, identity stack disruption, and CTV consolidation.

What was covered

  • Fox acquiring Roku for ~$22 billion (deal expected to close ~2027), with hosts noting Fox already owns Tubi (a free ad-supported streaming service) and could combine both to become a meaningful FAST (free ad-supported streaming TV) player. Concern raised that Roku selling ad inventory for competitor streamers may create channel-conflict post-acquisition.

  • Hightouch offering Publicis up to $1.2 billion for LiveRamp's identity assets alone — roughly half of Publicis's ~$2.2 billion total LiveRamp deal — with a one-week deadline. Hosts debate whether Hightouch's "composable CDP" (customer data platform) already replicates LiveRamp's identity layer or needs to acquire it to close the gap.

  • SpaceX IPO at ~$2 trillion valuation, followed by a ~$60 billion acquisition of Cursor (an AI coding application). Hosts frame this as a major bet on the AI application layer — distinct from foundational models — with implications for ad-tech M&A logic.

  • Pre-Cannes AI announcement wave: Horizon Media added agentic (AI-driven autonomous) buying to its platform; Xandr/Next opened AI agents across DSP and SSP (supply-side platform); DoubleVerify (DV) launched "Nora," a cognitive verification engine; Fox and Disney each launched AI-powered ad tools. Three of five announcements involved buyer/seller agent transactions.

  • The Trade Desk (TTD) and Publicis settled their fee dispute; Publicis is once again recommending TTD to clients. TTD also launched a travel/hospitality retail media network bundling Uber, Booking.com, United (Connective Media), and Marriott ("Riot Media").

  • Patrick Dolan (COO, OAAA) on OOH: 19 consecutive quarters of growth; $9.46 billion in U.S. revenue in 2024; digital OOH (DOOH) now 36% of total revenue; programmatic is a growing but still smaller slice of that 36%; AI is entering planning, workflow automation, and buyer/seller agent tooling.

Notable claims & predictions

  • Ari Paparo on Hightouch: "It's a little bit stunt M&A... but if you've got it all already, why do you want this $1.2 billion? You know, one-third of your total equity stack, basically, to buy this thing that you don't really need?" — implying Hightouch's claim to already replace LiveRamp may overstate its actual capability.

  • Paparo on Fox/Roku: "Anyone that's not part of one of the giant holdcos or giant tech companies [will] get acquired to fill in the gaps." — broad prediction for continued streaming consolidation.

  • Patrick Dolan (OAAA) on AI in OOH: "It reminds me very much of the early programmatic days... I remember going into a room and talking about data. The buyers were nodding their heads and nobody really understood what was going on. You can see where that ended up." — suggesting AI in OOH is early-stage but trajectory is transformational.

  • Dolan on retail OOH measurement advantage: Retailers can run controlled experiments — turn screens on in some stores, off in others — and measure direct sales lift (e.g., bottles of Pepsi sold), giving OOH a causal attribution edge over other media.

  • Paparo on SpaceX/Cursor: "The bet for a lot of ad tech is that ads and ad tech is an application layer for AI. This is a massive acquisition and a massive bet on the application layer... it might signal that there's more M&A to come across everything."

Names mentioned (from the watchlist

  • Fox — acquiring Roku for ~$22B; also announced agentic advertising capabilities pre-Cannes.
  • Roku — being acquired by Fox; noted vulnerability from Walmart/Vizio replacing Roku OS in white-label TVs.
  • The Trade Desk (TTD) — settled dispute with Publicis; launched travel/hospitality retail media network with Uber, Booking.com, United, Marriott.
  • Publicis — acquiring LiveRamp; received unsolicited $1.2B offer from Hightouch for identity assets specifically.
  • LiveRamp — identity/data onboarding asset being acquired by Publicis; Hightouch attempting to carve out identity layer.
  • DoubleVerify — launched "Nora," described as a cognitive/AI verification engine.
  • Disney — launched AI creative tool for SMBs (small-to-mid-size businesses), described as similar to Comcast's universal ad product.
  • Walmart — mentioned for Vizio acquisition enabling replacement of Roku OS in its private-label TVs, pressuring Roku's hardware distribution.
  • Amazon — noted as a top seller of Roku devices, creating strategic tension for Roku.
  • Databricks — announced an agentic CDP (customer data platform) in the same week as Hightouch, seen as competitive pressure on Hightouch's composable CDP positioning.
  • Snowflake — mentioned as an underlying data layer that Hightouch depends on, now potentially competitive.
  • Snap — launched Spectacles AR glasses at ~$2,000, contrasted unfavorably with Meta Ray-Bans on price and aesthetics.
  • ID5 — mentioned in context of building an independent identity stack post-LiveRamp acquisition.
  • Ari Paparo (Marketecture) — host; co-framing all segments.
  • Terry Kawaja / LUMA — not mentioned.
  • Arthur Sadoun (Publicis) — referenced indirectly through Publicis/LiveRamp deal discussion.

Why this matters for ad-tech operators

  • LiveRamp identity layer is in play. Hightouch's bid — even if it fails — signals that buy-side and data infrastructure players see LiveRamp's identity spine as separable and acquirable. Publishers and buyers relying on LiveRamp for data onboarding and audience connectivity should monitor whether the Publicis deal closes intact or is restructured; a fragmented outcome could disrupt existing data clean room (secure data collaboration environment) and identity workflows.

  • Fox + Roku creates a new CTV/FAST power block. Combined with Tubi, Fox would own a full stack: content, OS/hardware distribution, and ad sales. For buyers, this concentrates CTV inventory and could raise CPMs or restrict cross-platform data access. For rival streamers currently monetizing through Roku's ad platform, the acquisition introduces a direct conflict-of-interest risk.

  • Agentic ad buying is moving from demo to deployment. Three of five pre-Cannes AI announcements (Horizon, Fox, Xandr) specifically involve autonomous buyer/seller agents transacting media. OOH is now part of that conversation. Ad-tech operators building DSPs, SSPs, or planning tools should treat agent-to-agent transaction protocols as a near-term infrastructure requirement, not a roadmap item.

  • OOH programmatic is growing but measurement remains immature. At $9.46B total U.S. revenue with only 36% digital and programmatic a fraction of that, OOH remains small relative to digital display or CTV — but it's growing for 19 straight quarters. For buyers evaluating omnichannel attribution models (methods for crediting which ad drove a conversion), OOH's controlled in-store test-and-control methodology offers unusually clean causal measurement, particularly in retail media contexts.

Full analysis

Decision Council — Episode 178: Cannes Prep, OOH, and the Agentic Wave

Step 1 — Frame

This is a Cannes-eve grab-bag: a Fox/Roku megadeal, a hostile-feeling bid for LiveRamp's identity guts, and a flood of "agentic AI" announcements timed to the Croisette. The real question for ad-tech operators: which of these is a structural shift you have to build for, and which is Cannes theater you can safely ignore until the bill arrives?

  • Reversibility: Mixed. Fox/Roku and any LiveRamp carve-out are Type 1 (hard to reverse, multi-year). Adopting agentic buying tools is mostly Type 2 (easy to pilot, easy to drop).
  • What's actually being decided: Not "is AI real" — it's whether the plumbing of who-transacts-with-whom (identity layer, agent-to-agent protocols, CTV supply ownership) is being repriced right now.
  • Forcing function: Cannes week itself, plus the one-week Hightouch deadline on LiveRamp and the ~2027 Fox/Roku close.

Proceeding.

Step 2 — The Council

The Market Analyst. Three deals, one logic. Fox buying Roku (~$22B) is the holdco-or-get-acquired thesis playing out — anyone without scale gets absorbed to "fill in the gaps," as Paparo put it. Hightouch's $1.2B run at LiveRamp's identity assets — a third of its own equity for something it claims it already has — is either a genuine repricing of identity infrastructure or, more likely, a stunt to spook Publicis. SpaceX buying Cursor for ~$60B is the loud signal: capital is betting on the AI application layer, and ad-tech is positioning itself as exactly that layer. Plain version: the money is no longer flowing to who builds AI, but to who wraps it in something a customer pays for. Expect more M&A across the board through year-end.

The Skeptic. Count the announcements: Horizon, Xandr, DV's "Nora," Fox, Disney — five AI launches, conveniently the week buyers are drinking rosé. The load-bearing assumption is that "agentic" means anything yet. AdExchanger's own coverage features a vendor (Concord) whose pitch is less AI, and Digiday's second Cannes piece is literally about the unpaid bill behind the transformation talk. Hightouch claiming it already replicates LiveRamp while bidding $1.2B to buy it is the tell — the category is selling capability it hasn't shipped. Plain version: a lot of this is a press release wearing a product costume. The one substantive thing here is OOH's measurement story, and that got the least hype.

The Operator. Forget the strategy decks — what breaks Tuesday? Agent-to-agent transactions sound clean until your DSP's agent negotiates with an SSP's agent and nobody can reconstruct why a bid cleared at that price. Three of five announcements involve buyer/seller agents transacting; that's a reconciliation and brand-safety nightmare before it's a feature. The QP/DSP roadmap reality — onboarding "some manual steps remain," integrations still in flight — is the honest baseline everyone's working from. Meanwhile the Cursor offboarding standup shows even internal AI tooling churns on price and renewal dates. The gap between "agentic at Cannes" and "agentic in production" is 12–18 months, minimum.

The Customer / End User (the media buyer). Patrick Dolan's OOH segment is the only place a buyer hears something genuinely useful: retailers turn screens on in some stores, off in others, and count the actual bottles of Pepsi sold. That's clean causal measurement — the thing every other channel fakes with attribution models. Buyers don't want more agents; they want to know what worked. The Omnicom/Disney sequential-streaming deal in the reading list is the same instinct: solve a real viewer/buyer pain, not a demo. Plain version: buyers are being sold autopilot when what they asked for is a working speedometer.

The CFO. The Digiday "talk is about the bill" piece is the whole story. Every AI tool launched this week carries inference cost, integration cost, and retraining cost that nobody's amortizing publicly. Hightouch contemplating a third of its equity for an asset it may not need is a cautionary tale in capital discipline. The defensible spend right now: identity continuity (if your LiveRamp dependency gets carved up, your data workflows break) and measurement that pays back in proven lift — i.e., the OOH retail-test model. Agentic pilots should be cheap, time-boxed, and killable. Don't capitalize hype.

Step 3 — The Tensions

  1. Analyst vs. Skeptic on whether this is a repricing or a performance. Is the application-layer M&A wave (SpaceX/Cursor, Hightouch/LiveRamp) a real shift in where value sits — or capital and PR teams getting loud before Cannes? They genuinely part ways here.
  2. Operator vs. the announcement narrative on agentic readiness. The decks say "deploy now." The roadmaps and offboarding notes say the plumbing isn't done. If agents transact before reconciliation and brand-safety controls exist, the first scandal sets the category back a year.
  3. Hype budget vs. substance. The Customer and CFO agree the least-promoted story (OOH causal measurement) is the most useful, while the loudest stories (five agentic launches) are the least proven.

Step 4 — Synthesis

What this episode actually hinges on: whether "agentic" is an infrastructure shift you must build for now, or a marketing layer you pilot cheaply. The council leans toward pilot cheaply — the production gap is real (Operator), the bill is unaccounted-for (CFO), and the loudest claims are the least shipped (Skeptic). But two items are genuinely structural and Type 1: the LiveRamp identity layer being in play, and CTV supply concentrating under Fox/Roku. Those deserve real attention because they change the plumbing whether or not the AI hype is real.

De-risk this way: treat agent-to-agent buying as a time-boxed pilot, not a roadmap commitment; map your LiveRamp identity exposure now in case the Publicis deal gets restructured; and steal the OOH playbook — controlled on/off measurement — into any channel where you can run it.

Step 5 — The Prediction

The cleanest, most-grounded question the episode raises is the Hightouch/LiveRamp carve-out, with its hard one-week deadline.

Prediction: Publicis will close its acquisition of LiveRamp with the identity assets intact, rejecting Hightouch's carve-out bid, before Publicis's Q3 2026 earnings report.

Confidence: Medium — Hightouch's bid reads as a stunt; Publicis wants the identity spine whole.

Revisit by 2026-10-31: We're right if Publicis completes the LiveRamp deal without divesting the identity assets to Hightouch. We're wrong if Hightouch (or another party) successfully carves out LiveRamp's identity layer, or Publicis restructures the deal to exclude it.

The bid's own contradiction — paying a third of your equity for something you claim to already have — signals leverage play, not genuine intent. Publicis acquired LiveRamp precisely for that identity connectivity; surrendering it for half the price defeats the purpose. The one-week deadline is pressure theater, and pressure theater rarely closes billion-dollar carve-outs.

Listen to the episode →

Comments