Retail Media Under Pressure: the independent shrinks and the buying decision moves to the assistant
Vista Equity Partners and Quinti Capital made an offer to acquire Criteo for around USD 3.7 billion, more than a 50% premium over its market price. Criteo connects roughly 17,000 brands with about 225 retail media networks. If the deal closes, the asset moves into private equity hands. Bloomberg and ppc.land reported it in early July 2026.
In parallel, and with no apparent connection, Amazon blocked OpenAI's crawlers, keeping its catalog out of reach of the agents, and a few weeks ago it began buying ads in ChatGPT. Juozas Kaziukėnas of Marketplace Pulse documented it. Andy Jassy, Amazon's CEO, acknowledges that agents still show inaccurate prices and incomplete product information, and the company invests anyway.
The backdrop of the new channel: OpenAI opened self-serve ads to advertisers in the United States in May 2026 and added product feed ads in June. In LATAM they still can't be bought, and for Argentina there is no date. In retail media, the region is already a concentrated market: Mercado Ads dominates, Amazon Ads and Walmart Connect carry weight in Mexico, and the reference neutral player was, essentially, Criteo.
The Analysis
The two moves are directions of the same current. One is consolidation: an advertising-infrastructure asset moving from the public market to a fund. The other is migration: the part of the funnel where someone asks "what should I buy" starts to be answered by an assistant before a product search engine. The thread that ties them is not causal, it is structural. Both are moves over control of inventory and of purchase-intent data.
When private equity takes a company like this, the historical pattern is familiar: less transparency of results and adjustments to commercial terms. It is not inevitable, but it repeats often enough to treat as a base case rather than a surprise. For a brand running retail media, that turns a latent question into an operational one: how much of the investment runs on networks whose transparency and rules depend on a single owner. This is not an argument against operating retail media. It is knowing exactly who your investment depends on, because the day that owner changes its terms or closes off reporting, concentration stops being a market fact and becomes an account fact.
Amazon's move illuminates the logic of the other extreme, the walled garden. Proprietary data is not shared: that is why it blocks the crawler. But presence is bought where the user decides: that is why it pays to be in ChatGPT. That a company with the most-used product search engine in the world invests in a channel its own CEO calls imperfect says less about the channel's maturity than about its strategic value. Part of the buying decision is shifting. Not all of it, but the part that weighs most in early consideration. This connects with a broader shift, the erosion of traditional search against AI search, and with the mechanics of how ads are measured inside ChatGPT when there is still no attribution standard.
The Tensions
The most obvious objection is that tying a financial M&A to an ad purchase is forcing a narrative. It is a fair critique if read as cause and effect. It is not that. The connection is that both moves redistribute the same scarce resource: control over where and with what data the buying decision is made. Criteo redistributes it through ownership; Amazon, through the channel.
The second objection is geographic. If ChatGPT ads still can't be bought in LATAM and Criteo weighs less than Mercado Ads, why should it matter today? Because the work that puts a brand in order against these changes pays off in the channels it already uses, without depending on the new channel opening. A clean product feed, with up-to-date titles, prices and stock, improves performance in current channels and is also the entry requirement for the formats OpenAI already launched abroad. Solid server-side measurement serves today and serves more when there is traffic to attribute that does not behave like Google's or Meta's. Preparation is not betting on the channel too early. It is groundwork that is not wasted in any scenario.
The third tension is one of degree. None of this forces abandoning retail media or treating walled gardens as adversaries. The useful reading is not defensive, it is architectural: understanding which part of the investment depends on a third party's infrastructure and reporting, and which part of future demand will form inside an assistant that the region still can't buy from.
For Whom
This reading matters most to brands with an active presence in retail media networks outside Mercado, to e-commerce whose performance depends on the quality of a product feed, and to teams that already think about measurement beyond what each platform reports. For a small business, without meaningful retail media investment and without a catalog, it is industry context more than an action for this week. The difference lies in how much of the result depends today on platforms that are not controlled.
What We Don't Know
It is not confirmed that the Criteo offer will close, nor on what final terms. The pattern of private equity adjusting transparency and prices is a historical trend, not a certainty about this case. There is no date for ChatGPT Ads opening in LATAM. And the real magnitude of the buying decision shifting toward assistants is still measured with early data: the direction is clear, the size is not.
Sources
- Bloomberg — Vista Equity Partners and Quinti Capital offer for Criteo (July 2026)
- ppc.land — coverage of the deal and premium terms
- Digiday — independent retail media context
- Marketplace Pulse (Juozas Kaziukėnas) — Amazon blocks OpenAI's crawler and buys ChatGPT ads
- eMarketer — retail media and agentic commerce
- OpenAI — self-serve ads launch (May 2026) and product feed ads (June 2026)
This piece was produced with Zenda's editorial system, which combines AI-assisted analysis with human review. Data and sources were verified as of the publication date.