August 26, 2025
by Charlie Swift, General Manager, Adstra Services
DSP. MCP. AEO. RMN. DCP. The industry runs on TLAs, so we thought we’d check in on the state of media through five of them.
Each reflects a shift in how digital systems are being built, accessed, or governed. Some are more mature than others, but all serve as signs about where the industry is moving.
DSP: Demand-Side Platforms, Supply Path Optimization, and the New Supply Chain
The DSP market has consolidated. With MediaMath gone and The Trade Desk in a dominant position, buyers have fewer paths into premium inventory, particularly in CTV. Over the past year, Amazon’s DSP has emerged as a meaningful counterweight. Roku, Disney, and others are routing inventory through it, making it more attractive to buyers seeking alternatives.
Another TLA—SPO, or Supply Path Optimization—continues to shape how media is bought. Buyers are trying to reduce fees, cut through redundancy, and gain more transparency into how inventory is sourced. But SPO only matters if there are meaningful choices to optimize between. As DSPs and SSPs converge, and as Amazon gains ground, buyers are reexamining who sits between them and the media they want to buy and what value that intermediation adds.
The Trade Desk still holds a commanding position. But the shape of the competitive field is starting to shift.
MCP: Model Context Protocols and the Agentic Future
Model Context Protocols (MCPs) are emerging as a foundational component of agentic AI. These protocols connect AI agents to tools, memory systems, and environments. They enable agents to take action across systems and are quickly becoming critical to how agents actually operate.
This raises important questions around authorization, identity, and consent. What kind of identity system allows an agent to transact or make decisions on behalf of a user? How is consent granted, managed, or revoked? What level of visibility or control should a person have over the agent acting in their name?
The answers aren’t yet clear. But persistent identity, agent-level permissions, and new forms of validation will likely play a major role in how these systems evolve.
AEO: AI Engine Optimization and the New Front Door of the Internet
AI Engine Optimization (AEO) is redefining digital discovery. Tools like ChatGPT and Perplexity are the new front door of the internet, and are changing how people search for information, make decisions, and relate to brands overall. Unlike traditional search engines that produce a ranked list of results, these systems synthesize inputs to deliver a single conclusive output. That changes everything.
For marketers and content creators, this introduces a new set of unknowns. There is no results page, no clear ad auction, and no familiar ranking logic. The methods of influence, optimization, and monetization are still emerging.
Publishers face a related challenge. Their content is being summarized by LLMs and shown to users without guaranteed attribution or compensation. Some have entered licensing deals. Others are excluded entirely. The structure for rewarding content contribution remains unresolved. The long-term implications for publisher economics are significant.
RMN: Retail Media Networks and the Fragmentation Problem
Retail media networks hold valuable shopper data, but they are siloed. Each retailer—Walmart, Target, Kroger—operates independently. This makes it hard for buyers to get a comprehensive view across platforms or achieve efficient scale.
Third-party providers like Attain are building aggregation layers using finance and loyalty app data to provide cross-retailer insights. Clean rooms are also being used for collaboration, though most are controlled by platforms or intermediaries with their own interests.
There is precedent for solving this. Data co-ops in the catalog era enabled shared infrastructure across competitive parties. A similar approach could address current limitations in retail media. Whether through neutral platforms, shared clean rooms, or emerging intermediaries, the pressure to integrate retail data is increasing.
DCP: Data Collaboration Platforms and the Question of Control
Data collaboration has become a battleground. What began as neutral clean room infrastructure has become an extension of the commercial stack, owned and operated by agencies, platforms, or large publishers with their own interests to protect.
The result is less neutrality and more strategic leverage. Agencies are using their clean rooms to lock in clients. Publishers are building theirs to control monetization. Identity providers are embedding collaboration into broader offerings. Each move is defensible. But the effect is a narrowing of shared, unbiased space for data interoperability.
This is especially visible in measurement and attribution. If everyone grades their own homework, trust in outcomes becomes harder to maintain. Whether DCPs evolve into truly independent infrastructure or remain proprietary tools will determine how collaboration plays out over the next few years.
TLAs FTW
The best way to follow what’s shifting in digital is to pay attention to the acronyms. DSP, MCP, AEO, RMN, DCP—each reflects a deeper change in how media works, how data moves, and how decisions get made. The labels may be reductive, but the forces behind them are real.