Ad Tech Briefing: Amazon launches MCP server for agent-driven advertising

In practice, this allows agents using platforms such as ChatGPT or Claude to create campaigns, pull reports, and manage billing without custom integrations.

To recap, Model Context Protocol is an open standard for how AI systems communicate with external tools and AI agents, enabling agents to access Amazon Ads API functionality without requiring custom, one-off integrations, eliminating the need for multiple point-to-point connections and ongoing maintenance. 

This means advertisers and partners can use a single integration to connect custom-built agents or AI platforms such as Claude, ChatGPT, or Gemini. Once connected, the Amazon Ads MCP server allows agents to access individual Amazon Ads capabilities.

What’s more consequential is Amazon’s focus on workflow orchestration. Rather than exposing isolated capabilities, it bundles common ad operations — campaign launches, geographic expansion, reporting — into prebuilt tools that agents can execute end-to-end.

The announcement was made by Paula Despins, vp of measurement, Amazon Ads, where the company noted how “connectivity alone doesn’t guarantee reliable outcomes,” particularly when real-world ad tasks span dozens of steps.

For some, this move represents Amazon’ is betting’s bet that whoever standardizes agent execution early gets to define how advertising work actually gets done.

Amazon’s move lands squarely in the middle of a broader industry debate about how agentic advertising should actually work — and who gets to define the rules.

Initiatives such as the Ad Context Protocol, or AdCP, aim to create an “OpenRTB-like” standard for agent-driven media execution, enabling AI systems to buy, optimize, and manage ads across platforms without bespoke integrations. Proponents frame this as neutral, open plumbing for an agentic future; critics worry it risks entrenching existing power structures under a new technical banner.

By anchoring its MCP Server to Amazon Ads’ APIs and workflows, Amazon’s approach appears to be a more closed ecosystem, contrasting with AdCP’s cross-platform ambitions.

More broadly, there remain questions about governance, incentives, and enforcement of industry standards, with observers noting a key tension among industry bodies and significant debate over interoperability.

Meanwhile, platforms like Amazon are racing ahead to operationalize agentic advertising on their own terms — and setting expectations advertisers may soon demand elsewhere.

What we’ve heard

“Personally, I think it’s overhyped, they’re trying to essentially recreate programmatic, turning AdCP into the pipes, and someone’s going to control AdCP, which essentially replaces the idea as DSP and SSP… do you think that a lot of these big DSPs are just gonna kind of roll over?

— An anonymous source hints at the unseen tip of the iceberg that is the latest industry protocol over agentic advertising, which could cause tensions beneath the surface. 

Numbers to know

• $63 billion: The amount of ad spend wasted on invalid traffic on platforms, per the latest report from Lunio.  
• $1.5 billion: The amount sought in “personal punitive damages” from LG Electronics in court papers filed by Alphonso’s founders, according to AdExchanger. 
• 10%: The percentage reduction in corporate headcount by Amazon – it confirmed 16,000 job cuts the latest round of its now (almost-quarterly) sacrifice of headcount ahead of its February 15 earnings call – since October. 
• 15%: The percentage of headcount cut by Pinterest equating to 780 people, in its latest cull of staff ahead, equating to 780 people per its latest official statistics, ahead of its February 12 earnings call.  

What we’ve covered

Programmatic advertising on TV is set to increase this year

As streaming and CTV edge closer to absorbing more of linear TV’s share of budgets this year, programmatic advertising is set to account for a larger slice of the TV advertising pie.

Google’s forced AI opt out: what changes — and what doesn’t — for publishers

The U.K.’s competition watchdog has said publishers should be able to opt out of their content being used in Google’s AI Overviews without it affecting how they appear in search engine results.

Google said in response it is already “exploring updates to our controls to let sites specifically opt out of search generative AI features.”

What we’re reading

Microsoft to stop caching Prebid video files, leaving publishers with a major ad serving problem

AdExchanger’s James Hercher notes how Microsoft will stop caching Prebid video files on April 30, 2026, shortly after it shuts down the Xandr DSP, affecting how video ads serve for publishers using Prebid and Google Ad Manager. The change ends a free infrastructure service many rely on, meaning that without migration to paid cache services or alternative setups, video ads may fail to render and create costly ad-serving problems for publishers.  

PubMatic expands commercial leadership team to accelerate buy-side and publisher growth

Last week, PubMatic made what, on the face of it, were hum-drum appointment announcements, with Joseph Dressler and Bill McLaughlin now both svp, advertiser solutions, covering brands and agencies, respectively. The high-profile appointments buttressing CRO Kyle Dozeman’s team point to the ever-increasing conflict between traditional DSPs and SSPs, which has been brewing for some time. So, just how will this play out in 2026 

AI layoffs or ‘AI-washing’?

A Forrester report published in January argued, “Many companies announcing A.I.-related layoffs do not have mature, vetted A.I. applications ready to fill those roles, highlighting a trend of ‘A.I.-washing’ — attributing financially motivated cuts to future A.I. implementation.”

Google quietly tapped a  veteran Exec to revamp its mid-market machine

Google has elevated longtime exec Jitendra Kumar to oversee the tech giant’s relationship with critical mid-market agencies, writes Adweek’s Lauren Johnson.