Publicis to buy LiveRamp for $US2.2bn as AI race shifts to data pipes

Publicis Groupe has agreed to acquire LiveRamp for $US2.167 billion in cash, casting the deal as a way to accelerate data co-creation for smarter agents. The announcement reads like standard M&A. The transaction is a $US2.2 billion bet that enterprise AI competition is moving past model releases and into the identity, consent and data-sharing plumbing that determines what an agent can actually do in production.

The price is $US38.50 a share, a 29.8 per cent premium to LiveRamp’s 15 May close, according to Reuters’ market report. Publicis framed the deal around identity, data collaboration and connected media. For an enterprise technology audience, those layers are the infrastructure most organisations still need before AI software moves from demos into workflows that touch real customer information.

What Publicis is buying matters more than the valuation. LiveRamp’s core assets are software and data rails that let businesses match, share and activate information across partners without pouring raw customer records into every new tool. Enterprises trying to roll out copilots or autonomous agents keep hitting the same questions: who can access what, how do you match records across systems, and can anyone audit the result before the system acts on live data?

Arthur Sadoun, chairman and chief executive of Publicis Groupe, said the acquisition showed the company’s commitment to investing “ahead of market shifts”. Corporate boilerplate, but the move itself is decisive. Publicis is treating identity and data collaboration as core AI infrastructure, not a back-office tidy-up to be handled after the model layer is chosen.

Reuters said LiveRamp connects more than 25,000 publisher domains across 14 markets. Techmeme’s roundup described LiveRamp as a platform for sharing and building data sets and models that power agentic frameworks. Enterprise data plumbing rarely makes headlines, until an organisation wants AI touching production systems. Then it is the only conversation that matters.

Why the data layer matters

Identity resolution, consent management and partner data collaboration have been important for years. Cookies weakened. Privacy expectations tightened. The change is that vendors now want those same controls underneath agents that can recommend, target or trigger actions across several systems at once. If the data layer is messy or poorly permissioned, the agent layer turns risky within weeks. A company better known for advertising and media services is spending billions on governed data access — that is the story.

Australian enterprises have been circling the same problem. AI strategy discussions fixate on model choice. The harder, slower work is data readiness, privacy controls, integration. Publicis did not pitch LiveRamp as a media-buying trophy. It pitched the company as a data foundation for clients that want AI systems to move from chat interfaces into workflow and decision support.

Scott Howe, LiveRamp’s chief executive, said customers and partners had been the company’s “North Star” and that joining Publicis would give it greater resources and flexibility to scale. The quote does not prove the combined group will lead the next phase of enterprise AI. It does show where the leverage sits: in the pipes, not just the models.