Nobody can say with confidence what programmatic advertising will look like in 2030, and any piece that states it as fact should be read with suspicion. What can be done is to set out the plausible end states and the evidence that would point towards each. Three scenarios cover most of the range: an assisted auction, where agents stay inside today's platforms; a two-track market, where auctions and agent-negotiated deals run side by side; and a fragmented negotiation market, where agents trade but on incompatible terms. Which one emerges depends on two open questions: whether agent protocols interoperate, and whether buyer and seller come to share one record of what they agreed.

TL;DR. Programmatic in 2030 is better planned as three scenarios than one forecast. In the assisted auction scenario, AI agents mostly optimise bids inside existing DSPs and negotiation stays with people. In the two-track scenario, the open auction keeps the long tail while premium and direct inventory moves to agent negotiation under shared deal records. In the fragmented scenario, agents negotiate widely but across competing protocols and separate records, so disputes rise. Each scenario has observable signals, and several buyer actions make sense whichever one arrives.

Why plan programmatic around scenarios rather than forecasts?

Scenarios are more useful than forecasts because the forces shaping programmatic are not settled. IAB Tech Lab announced AAMP, its Agentic Advertising Management Protocols, in early 2026. AgenticAdvertising.org reports the first agent-to-agent media buy using AdCP, the Ad Context Protocol, in October 2025. Both are real, both are early, and their relationship to each other is still open.

A forecast asks a buyer to bet on one outcome. A scenario set asks a different question: which signals would show that one outcome is becoming more likely than another, and which decisions hold up in all of them. That is a more defensible basis for budget and contract decisions than any single prediction. The current state of the market is covered in What is the future of programmatic advertising?

What are the two uncertainties that separate the scenarios?

The first uncertainty is interoperability: whether agents built by different companies on different frameworks can transact with each other. AAMP builds agent capability onto existing IAB standards such as OpenRTB and the Deals API. AdCP defines campaign tasks that run over MCP or A2A. They overlap in purpose, and whether they converge, coexist or compete is unresolved.

The second uncertainty is the deal record: whether buyer and seller hold one shared statement of agreed terms. The AdCP documentation states that payments, settlement, disputes and makegoods remain separate integrations unless a specific AdCP contract covers them. AAMP standardises communication and offers but does not by itself provide a shared commercial record. Today that job falls to marketplaces and contracts.

Scenario 1: What does an assisted auction look like?

In the assisted auction scenario, AI agents become standard tools inside existing platforms but rarely negotiate with each other. They help plan, set bids, adjust pacing and write reports. Deals are still mostly agreed by people, and the open auction remains the default route for most spend.

This scenario fits a world where agent negotiation proves costly to govern, or where neither framework reaches broad adoption among sellers. Signals pointing towards it include: deal ID share in buyers' DSP reporting staying flat; few major sellers listing agents in the IAB Tech Lab Agent Registry; and RFP responses still arriving as bespoke documents rather than standard formats. A buyer in this scenario gains efficiency but little new accountability.

Scenario 2: What does a two-track market look like?

In the two-track scenario, the open auction continues to clear long-tail inventory while a growing share of premium, direct and packaged inventory is negotiated by buyer and seller agents. Negotiated deals are recorded at the moment of agreement in a form both parties hold, and humans approve at set thresholds.

This scenario needs both uncertainties to break the right way: agents interoperate well enough, and shared deal records become a normal contract term. Signals pointing towards it include: sellers returning RFP responses in IAB Tech Lab's OpenProposal format; rising deal ID share spread across many smaller deals; contracts naming a single record of agreed terms; and vendors stating which framework versions and deal types they support. The precedent is viewability, which moved from optional extra to routine contract term in the years after the Media Rating Council issued its viewable impression guidelines in 2014.

Scenario 3: What does a fragmented negotiation market look like?

In the fragmented scenario, agent negotiation takes off, but on separate tracks. Buyers need different integrations for different seller groups, each side keeps its own record of what was agreed, and reconciliation after delivery becomes harder rather than easier because there are more deals with more terms.

This scenario arrives if agent capability spreads faster than shared governance. Signals pointing towards it include: vendors supporting one framework and dismissing the other; agent-agreed deals growing while contracts stay silent on who holds the record; and makegood disputes rising with negotiated deal volume. For example, a hypothetical buyer running 300 small agent-negotiated deals a quarter, each with five or six terms, faces far more reconciliation points than one running ten large insertion orders, unless both sides already hold the same record.

Which signals should a buyer watch each quarter?

Five indicators separate the scenarios and can be checked from data a buyer already has or can access publicly:

Deal ID versus open auction share of spend, by inventory tier, from DSP reporting.
The share of RFP responses that arrive in a standard format such as OpenProposal.
Agent Registry listings among the buyer's top sellers.
Whether new contracts name one record of agreed terms that both parties hold.
The number and cause of makegood disputes on negotiated deals.

Flat readings on the first three point to Scenario 1. Movement on all five points to Scenario 2. Movement on the first three with no change on the last two points to Scenario 3.

What should buyers do now that works in every scenario?

Three actions hold up whichever scenario arrives. Write mandates that state what an agent may do alone and which commitments need human approval, as described in how to write a media buying mandate. Ask every vendor which framework it supports, at which version, for which deal types. Require a record of agreed terms that the seller holds identically, retrievable by either side.

That third requirement is the one Alkimi is designed around. As a neutral marketplace working within IAB Tech Lab's AAMP framework, with its agent listed in the Agent Registry, it has buyer and seller agents agree terms on a DealSheet that is timestamped at agreement and held by both parties, and people keep approval rights over each deal.

Frequently asked questions

What will programmatic advertising look like in 2030?

No single answer is defensible. Three scenarios cover the range: an assisted auction, a two-track market of auctions and agent-negotiated deals, and a fragmented negotiation market. Observable signals show which is becoming more likely.

Will the open auction still exist in 2030?

All three scenarios assume it does. The open auction prices long-tail impressions at a speed and scale that negotiation does not match.

What decides which scenario happens?

Two uncertainties: whether agent protocols such as AAMP and AdCP interoperate, and whether buyer and seller come to share one record of agreed terms.

What is the most useful signal to watch?

Deal ID share of spend by inventory tier in a buyer's own DSP reporting, read alongside whether contracts name a shared record of agreed terms.

What can buyers do now without betting on one scenario?

Write clear agent mandates with approval thresholds, ask vendors for framework versions and supported deal types, and require a record of agreed terms both parties hold.

What Could Programmatic Look Like in 2030? Three Scenarios and the Signals to Watch