Xandr Invest DSP Sunsets
Today I find myself sitting in a café on Terrigal Beach, reflecting on the close of a chapter. In less than a week, (28th February, 2026) is when the long expected will finally happen – Microsoft is officially saying goodbye to Xandr Invest DSP, bringing an end to a whole decade long story that’s defined Gourmet Ads and the rest of the programmatic industry.
For me, it all started way back in 2012 when I took the leap and signed a contract with AppNexus. The ups and downs of changing ownership, new strategies, the constant evolution of technology and the cutthroat competition have all played a part in this journey – and we’ve built a fair bit of our business around it. As the DSP era winds its way down, it feels like the right time to think about what this journey’s meant to me, what lessons I can take away from it and what comes next for anyone working in programmatic – whether you’re a buyer, a publisher looking to sell, or someone who works deep in the programmatic ecosystem.
In the early days, I was always staggered by the level of innovation coming out of AppNexus HQ on 23rd Street in New York. There was an energy in that building that I have never experienced. It always seemed that the product ideas moved quickly. Engineers, data scientists and commercial teams were tightly aligned. I always felt that Appnexus genuinely believed it could reshape how media was bought and sold. For the most part it did.
The Birth of AppNexus – A True Pioneer in Programmatic Tech
AppNexus, before it was known as Xandr Invest, it was actually called ‘Console’. It blossomed into one of the first real Demand Side Platforms to actually give companies like Gourmet Ads powerful API access. It was a huge deal at the time because most other DSPs just didn’t offer this sort of thing. It was like the key to the kingdom – we could build just about anything we wanted on top of the platform and that felt seriously exciting.
When it hit the scene, AppNexus had a lot of firsts, it was among the first to support video buying, it was one of the first to introduce guaranteed 100% viewability and was an early mover in the whole viewability targeting space. Plus, they were a leader in performance based buying which allowed advertisers to bid on metrics like cost per click or cost per acquisition. These innovations really started to change how we looked at digital advertisng, forcing us to get creative and come up with new ways to deliver results for our clients.
Under Brian O’Kelly’s leadership, AppNexus was a true trailblazer. At AppNexus Summits in New York and London, Brian would constantly challenge partner businesses to try new things – it was a real challenge but also an incredible opportunity for growth. Of course, with innovation comes some speed bumps along the way, and there were definitely some missteps in AppNexus’ early days. I remember one afternoon at AppNexus HQ in New York with Brian and one of the data science team going over examples of the bad players on the platform. Back then the fraudulent activity was a real issue for everyone. It was clear to me that that data science was a key part of the solution. The company invested big (albiet a bit too late) to make the platform cleaner. AppNexus wasn’t perfect, bad networks and fraudsters gave the platform a bit of a rough name. But by the time it was sold to AT&T, a lot of those problems had been sorted out and the platform had been cleaned up.
The Telco Acquisitions and the AT&T Era
In mid 2018, it was announced that AT&T would acquire AppNexus. I wasnt thrilled, but at the time, I spoke with a few of the executive team who were genuinely excited about the opportunity. However, I remained cautious. I had seen other DSP acquisitions by telcos and the results were often underwhelming. Telstra’s acquisition of Videology in Australia is one example that later became a write off. Verizon’s purchase of Yahoo’s core internet business later also became a cautionary tale about telcos buying digital platforms.
One of the biggest selling points during the acquisition was the promise of AT&T’s data and identifiers becoming the major asset of the platform. The narrative was compelling. With access to insights from more than 170 million AT&T subscribers, the pitch was the DSP would have a unique identity graph and deterministic data advantage that no independent DSP could match. This was positioned as the hook that would bring major advertisers into the ecosystem. A powerful, telco driven identity layer combined with premium TV and digital inventory sounded unbeatable.
But in reality, that promise never fully materialised in the way many expected. The data was there, but activating it at scale across the open web, in a way that felt seamless and globally relevant, proved far more complex. Privacy concerns, regulatory scrutiny, technical hurdles and shifting market conditions diluted what was meant to be the platform’s defining advantage. The much anticipated data moat never quite became the growth engine it was forecast to be.
AT&T clearly wanted to put its own stamp on the acquisition, while distancing the business from some of the historical AppNexus quality issues. The AppNexus brand was retired and the business was reintroduced as Xandr, a nod to Alexander Graham Bell and AT&T’s heritage. Strategically, it was meant to signal a new chapter and anchor the platform within AT&T’s legacy.
However, the rebrand and shifting priorities coincided with the departure of many talented product and account leaders. Much of the deep programmatic expertise that had defined AppNexus walked out the door during this period. That loss of institutional knowledge, particularly on the technical and buy side innovation front, further slowed progress and made it increasingly difficult to maintain a clear, consistent product roadmap.
Add to that, internal AT&T priorities overshadowed a broader market focus. There were long periods where clarity of direction was lacking, and innovation felt like an after thought. I think it was 2019 when it felt like nothing happened as 90% of the development seemed focused internally, rather than on the wider customer base. Everything increasingly felt focused on WarnerMedia. We were actively encouraged to lean into the supposed synergies. With brands like Food Network, HGTV and other lifestyle properties sitting adjacent to our own food and health verticals, the story sounded compelling. The pitch was that there would be deeper integrations, premium inventory access and meaningful data alignment between telco data, television assets and digital scale. For businesses like ours, that proximity felt like an opportunity. But in practice, very little materialised. The conversations were there. The strategic narrative was there. The commercial execution never really followed. That lost momentum allowed rivals like Trade Desk, Google’s DV360 and to a minor extent Amazon DSP to take significant market share, leaving Xandr partners struggling to keep pace.
The Microsoft Reset
When the sale of Xandr to Microsoft was announced in December 2021, I was genuinely elated.
There had always been rumours that Microsoft was the under bidder during the original AppNexus sale process to AT&T. I often wonder what might have happened if it had gone directly to Microsoft back then. Microsoft had maintained a long relationship with AppNexus from the early days and there was already technical alignment. There was already mutual respect. In many ways, it felt like unfinished business finally being resolved.
By the time Microsoft acquired Xandr, the platform had struggled for identity. On the buy side especially, competitors like Trade Desk and DV360 had steadily taken share. Momentum had shifted. Confidence in the roadmap had wavered.
What Microsoft brought was something different. Not a telco mindset. Not a media conglomerate agenda. But a product and engineering culture.
The teams that remained were a natural fit. Many of the engineers and product leaders inside Xandr were always closer in DNA to a technology company than to a telecom operator. Under Microsoft, that alignment became obvious. Development cycles accelerated. Ideas moved from concept to beta to release faster. There was clarity around ecosystem positioning rather than forcing Xandr to be the centre of the universe.
Importantly, Microsoft did not try to preserve Xandr as a standalone identity indefinitely. It made the strategic decision to integrate the assets into a broader Microsoft Advertising ecosystem. That was the right call. Running parallel buy side platforms long term would have made little sense. Consolidation was inevitable.
The recent rollout of the new Both Monetize and Curate homepages, deeply integrated with Microsoft Copilot, is a tangible example of this shift. Snapshot dashboards powered by AI. Intelligent summaries. Practical application of machine learning rather than just talking about it. That is product velocity.
With Invest being sunset, I expect even faster innovation on Monetize and Curate. Focus drives progress. And Microsoft is clearly focused.
Why Microsoft made the call, and what comes next
I can see why Microsoft leadership made this decision.
They already had Microsoft Advertising, a powerful self service buy side platform built on search and keywords, that has steadily been gaining features which feel increasingly aligned with a DSP. Audience layering. Native placements. Video. AI driven optimisation. The line between search and programmatic is blurring.
Maintaining two separate buy side technology stacks long term would have been cumbersome. Two engineering roadmaps. Two identity strategies. Two user interfaces. Two competing visions of what a should be. At some point, consolidation becomes not just logical but necessary.
From a leadership perspective, simplifying the buy side into a single, evolving Microsoft Advertising ecosystem allows focus. It allows AI, Copilot and identity investments to sit in one place rather than being fragmented across platforms. It allows Microsoft to compete more directly with Google, Amazon and Yahoo with a unified story.
And let’s be clear, the DSP market is brutally competitive. The Trade Desk has built enormous momentum. DV360 is deeply embedded in agency workflows. It has the unique inventory of Youtube. Amazon continues to tighten the loop between media and commerce. In that environment, running a legacy DSP alongside a growing self service ad stack may have felt like splitting resources at the wrong time.
So while there is nostalgia attached to Xandr Invest, strategically, I understand the call.
My view is that Microsoft is not exiting the buy side of programmatic. It is redefining it within its own ecosystem.
The keyword driven interface of Microsoft Advertising is already evolving beyond traditional search. Add Copilot. Add first party signals from LinkedIn, Xbox, Windows and Bing. Add retail integrations and native inventory. The foundations are already there for something that looks and behaves like a modern DSP, even if it is not branded as one.
The next logical step is AI led media buying.
Not AI as a reporting layer. Not AI as a recommendation widget. But AI as the operating system for campaign construction, optimisation and budget allocation.
Instead of building line items, uploading spreadsheets and manually stitching audiences together, buyers will increasingly brief outcomes. The platform will determine mix. It will test channels. It will reallocate spend in real time. It will optimise toward business signals rather than media metrics.
That future aligns far more naturally with a unified Microsoft stack than with a legacy standalone DSP.
If Copilot becomes embedded at the planning stage, connected to first party identity graphs and supply side integrations, the buying experience changes completely. It becomes less about toggling settings and more about instructing an intelligent system.
And that is where I believe Microsoft is heading.
The irony is that many of the early experiments at AppNexus, decision trees (Bonsai), Automated Optimisation, logic driven media planning, were primitive versions of what AI buying is now becoming. In many ways, the industry is circling back to ideas that were ahead of their time.
Only now, the technology has caught up…. or is fast catching up.
