Criteo Is Getting Ready to Power AI Shopping While Everyone Else Debates It
Criteo‘s move to Luxembourg isn’t the story. The quiet infrastructure play behind agentic commerce is.
I was juggling two school runs in the rain and an 8 week old puppy’s obsession with biting my face when I saw the Criteo announcements popping up.
Three separate pieces of news. All dropped in the first week of January. All quietly telling us where commerce media actually thinks it’s heading... so I’ve broken it down here for you.
Let’s start with the corporate housekeeping that’s far more revealing than it sounds.
So... on 7 January 2026, Criteo confirmed it’s moving its legal domicile from France to Luxembourg, with a shareholder vote scheduled for 27th February.
It’s also scrapping the American Depositary Shares structure... which is basically the corporate equivalent of having someone else hold your wallet and charge you for the privilege, and listing ordinary shares directly on Nasdaq instead.
Cleaner, simpler, and cheaper... assuming shareholders sign it off by Q3 2026.
The official line from board chairman Rik van der Kooi is that this will “streamline our corporate structure, enhance our capital management flexibility and align our capital markets presence with our long-term strategic ambitions.”
Translation: they want more room to manoeuvre in a market that’s fragmenting faster than anyone predicted, and France’s corporate structure wasn’t giving them the agility they needed.
Thing is, nobody redomiciles to Luxembourg by accident.
It’s where companies go once scale, shareholders, and reality arrive. Less theatre, more spreadsheets, and lawyers who don’t flinch when the numbers get big.
Luxembourg offers capital flexibility, lighter regulatory friction, and a more investor-friendly environment. It’s the sort of move companies make when they’re preparing for either aggressive M&A or significant restructuring.
Either way, it signals that Criteo thinks commerce media is about to get messy!
But the bit that caught my attention wasn’t the Luxembourg move tbh...
It’s what Criteo published on the same day about where commerce is actually heading.
Agentic commerce isn’t going to replace existing shopping behaviours. It’s going to be incremental.
CEO Michael Komasinski points to Ark Invest’s prediction that 25% of ecommerce will be agentically driven by 2030. That sounds dramatic until you realise it only translates to just over 5% of total retail sales.
This is the bit most people miss when they start hyperventilating about AI agents buying everything for us.
Mobile didn’t kill desktop. Social commerce didn’t kill ecommerce. Every new channel adds surface area. It doesn’t consolidate it.
Agentic commerce follows the same pattern. It creates new opportunities by reducing friction rather than cannibalising what already exists.
That’s a far more realistic read of consumer behaviour than the “AI will autonomously manage all your shopping” narrative that some VCs are still flogging.
The data backs this up.
Criteo surveyed 10,000 shoppers globally in September 2025 and found that 40% of U.S. shoppers now use agentic shopping assistants regularly for product research. But 96% of them still use other channels along the way, including search engines, social platforms, and brand or retailer sites.
No displacement. Just more fragmentation.
Michael also dropped a detailed piece on agentic commerce in Decemeber that’s worth reading in full because it’s one of the more grounded takes I’ve seen from a platform executive... I’ll put the link in the comments.
Which brings us to the real tension in commerce media right now...discoverability across an increasingly fragmented ecosystem.
Komasinski argues that products need to be findable wherever consumers initiate queries, across retailer front ends, social platforms, LLM environments, and emerging AI-driven interfaces.
That’s not trivial.
It requires structured commerce data at scale.
Real-time inventory.
Accurate pricing.
Detailed product attributes.
Unified fulfilment.
Most LLM platforms simply don’t have this yet, which is why conversion performance is still poor.
He even admits it!
He describes spending nearly an hour inside an LLM platform trying to find puncture-proof city bike tyres, only to hit broken links, discontinued models, and incomplete product information. Eventually, he gave up and went to a local bike shop.
That’s agentic commerce today... impressive reasoning. Weak execution.
Even OpenAI ’s own analysis puts ChatGPT shopping research accuracy at around 64%. When you’re wrong over a third of the time, you’re not building trust. You’re burning it.
This is where Criteo’s positioning becomes genuinely interesting.
They’re not trying to build consumer-facing AI shopping agents. They’re betting on being the data and infrastructure layer that powers everyone else’s agentic experiences.
They already sit on 720 million daily active users.
4.5 billion SKUs in their Universal Product Catalog
Over one trillion dollars in annual ecommerce transaction data
That scale creates a continuously refreshed commerce intelligence layer. In a world where LLMs can reason brilliantly but can’t reliably tell you what’s in stock or how much it costs, structured data becomes the moat.
Retail media is the other half of this story.
Criteo’s President of Retail Media, Sherry Smith , published the company’s 2026 outlook on the same day.
And again, it’s unusually honest.
1. Self-service tools for brands and agencies finally have to work.
The current reality of juggling multiple retailer platforms with inconsistent data structures and reporting logic isn’t sustainable.
Criteo’s pushing towards unified execution, flexible measurement, and AI-powered optimisation without opaque black boxes.
2. Retailers need to stop managing ad revenue in silos.
Criteo is building software to help retailers align merchandising, media, and shopper experience goals through dynamic layouts, personalisation, and clearer control over the full journey.
This directly addresses a real, genuine pain point.
Too many retailers optimise ad yield independently from merchandising outcomes, which kills incrementality and long-term value.
3. Agentic AI will redefine discovery as shoppers shift from keywords to conversations.
That means sponsored placements inside retailer-owned assistants, AI-integrated retail apps, and commerce data feeds powering conversational interfaces. It also means monetisation inevitably shifting towards advertising.
Google is already showing ads inside AI Mode search.
ChatGPT has roughly 5% of users paying for subscriptions, which leaves advertising as the most obvious growth lever.
Advertising remains the most scalable digital monetisation model we have. In the GenAI era, it just rewards data quality, transparency, and relevance far more aggressively.
So what does this actually mean if you’re running campaigns?
The Luxembourg move won’t affect day-to-day operations in the short term, but it signals Criteo is preparing for bigger strategic shifts.
The agentic commerce positioning matters more.
If Criteo succeeds in becoming the structured commerce data layer behind AI shopping across platforms, it becomes essential infrastructure. If it fails, it risks being just another retargeting platform watching the ecosystem fragment around it.
The retail media roadmap targets genuine pain points, but the industry has heard self-service promises before.
The real test is whether Criteo’s unified approach actually reduces operational friction or simply adds another abstraction layer.
One thing is clear though... commerce media is fracturing across more surfaces than anyone expected even two years ago.
Retailer-owned assistants like Amazon’s Rufus saw sessions resulting in a purchase grow over 100% during Black Friday and Cyber Monday, compared to around 20% growth for non-assistant sessions.
That’s not noise. That’s a signal...
Conversational interfaces are creating incremental demand.
The platforms that control the data feeding those experiences are positioning themselves as the plumbing everything else depends on.
Whether Criteo pulls this off comes down to execution, not vision.
The vision is coherent.
Be the commerce data backbone for agentic shopping.
Unify retail media workflows across fragmented networks.
Monetise through scalable advertising embedded in conversational interfaces.
The execution requires global data integration at scale, genuine interoperability across retailers and LLM platforms, and accuracy north of 95% ... IF... consumers are ever going to trust these systems.
Oh and if you actually want to understand how this machine works... I literally wrote the book on it called WTF IS PROGRAMMATIC? (it’s on Amazon, obvs...)
This took me ages to research and write. All I ask is that you hit subscribe above^ if you found value.
k, thanks, bye