Containerized RTB and How It Shifts the Programmatic Space

By Dan Jaye, CTO

The Google AdEx privacy settlement and its introduction of RTB Control made one thing clear: identity enforcement is moving into the core pipes of programmatic. When consumers can opt out and strip identifiers at the auction level, the open bidstream becomes far less reliable. That’s why I believe the next wave of disruption won’t come from more workarounds—it will come from rethinking where decisions happen. And that’s where containerized RTB fits in.

When I talk about containerized RTB, I mean moving bidding logic and enrichment into the exchange environment so bid decisions happen right next to the auction. This isn’t about shifting auctions to the publisher side; header bidding has already shown how messy and duplicative that can get. The real promise lies in exchange-side containers, where the full set of impression signals is instantly available, latency is reduced, and publishers can still apply guardrails for secure data sharing.

Instead of shuffling bloated OpenRTB payloads back and forth (often missing critical fields anyway), containerized RTB lets us run the process inside the exchange. That means faster decisions, richer data, and a more governed way to handle signals.

Why It Looks Promising

There are three big reasons I’m excited about containerized RTB on the exchange side:

• Better signals – OpenRTB simply doesn’t carry enough data for the most informed decisions. Exchange-side containers expose richer impression context.

   

• Lower latency and cost – Cutting out round-trips between SSPs and DSPs removes network drag and bandwidth costs.

   

• Smarter governance – Running containers inside exchanges enables new guardrails for how data is shared, reducing risk.

Put together, this could mean smarter decisions, faster – and with more relevant signals flowing to the models that need them.

Where the Friction Is

The biggest challenge I see is coordinating across multiple exchanges. Once every SSP spins up its own container, frequency management and deduplication get messy. We already see this problem with header bidding, where the same inventory shows up in multiple places and buyers risk bidding against themselves. Without a fix, containers could make that problem worse.

Another limitation: many current pilots look more like sandboxed execution environments than true open compute fabrics. Think Snowpark, not Docker. You can bring in your model as an executable, but you can’t yet bring in buy-side data freely. Until that changes, the bigger promises of containerized RTB will remain constrained. 

Fewer Layers, Not More

I strongly believe marketplaces exist to simplify complexity, not add to it. Done right, containerized RTB should reduce layers, not pile on new ones. I can picture independent providers making it simple for brands to “sponsor a container” and connect more directly with SSPs. That’s the efficiency story that resonates: fewer intermediaries, not more.

What It Means for DSPs

Exchange-side containers also raise some hard questions for DSPs. If more decision-making occurs within exchange environments, some of the traditional DSP roles begin to shift. Agencies are already building meta-DSP tooling to orchestrate across platforms. In a containerized future, those orchestration layers may prove more valuable than traditional DSP stacks.

I’ve even wondered aloud about whether this is the “death knell for DSPs” moment, but it’s too early to call. What’s clear is that the value chain will change.

Where It Fits With AI

Containerized RTB also lines up with the way we’re starting to think about GenAI in advertising. Code and models generated upstream can be deployed as containers that run close to the auction. That means reduced latency, richer exchange-side signals, and near real-time model adaptation. It’s a natural fit for the AI-driven workflows now taking shape.

Where I Think This Shakes Out

• It’s still early days, with plenty of standards and plumbing to solve.

   

• Multi-exchange coordination is the most significant risk: without it, we risk cannibalizing ourselves.

   

• The most likely path is containers gaining traction alongside deal IDs, PAIRs, and private relationships.

   

Strategically, this isn’t about shifting auctions to publishers. Header bidding showed the limitations there. It’s about smarter signals at the exchange, fewer layers in the chain, and agency-led orchestration on top.

I think containerized RTB in my view, is the logical next step in programmatic’s evolution, and it may be exactly the structural shift needed to adapt to the identity shake-up already underway.