Microsoft Advertising Closing Xandr DSP, Layoffs Pending

Summary: Microsoft Advertising will sunset its DSP, Microsoft Invest (formerly known as Xandr) by February 28, 2026, per a company blog post. Microsoft’s embrace of AI is behind the decision; the company says the demand-side platform (DSP) model doesn’t align with its future vision of "more private and personalized advertising experiences for a conversational and agentic world." Layoffs are part of the move. Microsoft is consolidating its buy-side capabilities into the into the Microsoft Advertising Platform, a more simple, self-service buying tool (not too dissimilar from Google Ads and Meta Ads Manager). As part of its strategic shift to AI, Microsoft has integrated generative AI into its ad tools with Copilot.  Microsoft will continue to provide access to its ad inventory through third-party DSPs and will continue to prioritize its sell-side tech for publishers Microsoft Curate and Microsoft Monetize.  The Xandr DSP has a notable history, originating as AppNexus before being acquired by AT&T in 2018 and rebranded as Xandr. Microsoft then acquired Xandr from AT&T in 2021.

Opinion: When Microsoft cites "a conversational and agentic world" as incompatible with "the industry's current DSP model," they're essentially declaring the death of complex programmatic UIs. The future of media buying won't involve humans meticulously adjusting bid parameters in DSP dashboards. It will be AI agents having conversations with other AI agents to execute campaigns.

The economics make this pivot inevitable. Transparent DSPs operate on razor-thin margins with high operational costs and fierce competition. Meanwhile, black-box solutions like Google's Performance Max and Meta's Advantage+ (and the soon-to-be-launched Microsoft PerformanceAdvantageMax+++Turbo) already deliver fantastic performance with minimal human intervention. Because their promise is performance and ease-of-use, they are able to justify obscuring cost and taking high margins. Microsoft is simply the first major player with the courage to admit that continuing to invest in traditional DSP infrastructure makes little business sense when the entire industry is racing toward AI-powered automation. (To be fair, it's convenient courage, as Xandr has been losing market share and clout in the DSP sector for a while.)

This decision becomes even clearer when you consider Microsoft's booming $20B ad business—much of it driven by LinkedIn's walled garden. Why pour resources into a struggling open web DSP when your fastest-growing ad revenue streams don't even rely on it? LinkedIn's success demonstrates that controlled environments with proprietary data and simplified buying interfaces are where the real money is. 

What's telling is Microsoft's decision to retain their publisher-focused tools like Curate and Monetize, which will enable them to continue monetizing the open web using their data. This signals where the real value lies in tomorrow's ecosystem: controlling premium inventory access (supply) rather than facilitating its purchase (demand). In a post-cookie, agentic-AI world, where third-party data is less valuable and algos / UIs become increasingly commoditized, the moat comes from being the gatekeeper to quality inventory and first-party data, not necessarily the buying layer.

For the industry, this temporarily accelerates consolidation around the remaining major DSPs—The Trade Desk, Amazon, Google DV360, Yahoo, and other incumbents will absorb Invest’s displaced enterprise business. But the bigger story is how quickly we're returning to black-box buying models that DSPs were supposed to replace—just with "AI" as the new justification for opacity. 

Buyers, beware!

That's it for this week. Get the full Newsletter here: https://uof.digital/newsletters/

THIS WEEK'S CRF (CUTTING ROOM FLOOR) MEMES:

FOUND IN THE WILD-