An Interview with Chalice’s Adam Heimlich...
...on Ad Tech's Two Architectures. A conversation on why advertising’s demand side just split in two
Published in partnership with Chalice AI
This publication opens a series arguing that media services are splitting into two businesses with different architectures: pooled, platform-driven decisioning for the median advertiser, and per-advertiser decisioning for the enterprise. Madison and Wall puts the paper’s argument to its author, Adam Heimlich, CEO of Chalice.
About Chalice: Chalice builds custom AI-driven allocation models that let individual enterprise advertisers run their own value function above platforms like The Trade Desk, YouTube, and Meta — rather than accepting the pooled, one-size-fits-all optimization those platforms build for the median advertiser. The goal is to turn ad spend from an expense platforms manage on autopilot into an investment each brand directs against its own definition of value. Learn more at www.chalice.ai.
Two acquisitions anchor your recent piece: Publicis buying LiveRamp and Walmart agreeing to pay up to $1.4 billion for Vibe.co. Why do two M&A deals in the same quarter tell you where the whole industry is heading?
ADAM HEIMLICH: Because they’re opposite bets, and they’re being made by two of the most sophisticated buyers in the business at the same time. The network side of advertising is consolidating into pooled, standardized infrastructure built for the median advertiser. The largest and most sophisticated advertisers are going the other way — building decisioning they own and control, and routing around the pooled stacks entirely. The demand side used to be one thing. Now it’s two.
That’s a strong claim — “the most consequential media decision a brand will make this decade.” Why put it in those terms?
ADAM HEIMLICH: Because which side you’re on determines who owns the decision for the next decade, not just the next campaign. Pick pooled when you don’t have the volume, the data, or the in-house capability to justify anything bespoke — a well-run pooled system will serve you better and cheaper than you could serve yourself. Pick enterprise when you have the scale to build your own value function, and staying pooled just means handing your economics to infrastructure optimized for someone else’s median. Get that wrong in either direction and you’re paying for the wrong architecture for years.
Before you lay out the new split, you spend real space on a decade of failed “stack” plays — Microsoft, AT&T, Verizon. Why start there?
ADAM HEIMLICH: Because the industry already ran this experiment once, at a cost of tens of billions of dollars, and it failed. The 2010s produced a graveyard of attempted stack plays. Microsoft bought aQuantive for $6.3 billion and wrote off essentially all of it. AT&T assembled AppNexus into a coherent thesis on paper that carrier culture smothered, then sold it to Microsoft for parts. Verizon combined AOL and Yahoo into “Oath,” took a $4.6 billion write-down, and sold the wreckage to Apollo. Every acquirer believed that owning the plumbing — ad servers, exchanges, data platforms — would manufacture a durable advantage like Google’s or Meta’s. Assembling Google’s components does not create Google’s advantage.
So what did those acquirers actually lack that Google and Meta had?
ADAM HEIMLICH: Pipelines to create and deploy billions of predictions in real time, in a loop where the same system serves the ad, observes the result, and feeds the next prediction. Google and Meta spent years and billions building that around their own media properties. The carriers had real data — downloads, purchases, location, traffic. What they lacked was the capability to refine that data into accurate bid-time valuations at auction scale. Even Microsoft, with real engineering behind it, found that buying an ad-serving stack didn’t deliver a prediction engine. Owning the boxes is not the same as owning the ability to price what flows through them. The new split is precisely about who owns that decision.
Walk us through your logic on each side — Walmart buying Vibe versus Publicis buying LiveRamp. You call them opposite bets.
ADAM HEIMLICH: Walmart buying Vibe is a smart platform play. Vibe serves mid-sized advertisers who’ve never bought a TV spot and want to launch a campaign from a laptop. Bolt that onto Vizio’s hardware footprint and Walmart’s purchase data, and you get standardized, self-serve, closed-loop advertising at enormous scale. For the median advertiser, pooled is the right answer — they lack the volume, the data, and the in-house capability to justify anything bespoke.
And Publicis buying LiveRamp?
ADAM HEIMLICH: That’s a hedge, not a network play. Publicis didn’t need LiveRamp’s technology to compete on the network side — it already owns Epsilon, a complete pooled stack serving its agencies at scale. What it wanted was LiveRamp’s position: the neutral layer through which advertiser data flows to everyone else. Sophisticated advertisers have begun pursuing custom models and per-advertiser decisioning that no agency or platform can read. By acquiring the neutral translation layer, Publicis is trying to stay relevant to both sides of the split — the legacy agency business below the sophistication line, and the independent-optimization market above it. The largest incumbent in the business is paying a premium because it sees the split coming. The whales are leaving for an architecture of their own.
You reach outside advertising entirely for the central analogy — institutional trading and prime brokerage. Why does that history apply here?
ADAM HEIMLICH: Brands with nine-figure ad budgets used to assume the platforms hold all the leverage. The history of financial markets says otherwise. Aggregator power rests on the condition that no individual buyer matters too much. When a small number of sophisticated buyers represent a disproportionate share of volume, that condition fails — a large buyer can credibly threaten to route around the aggregator, and the aggregator has to accommodate it. For decades, large buy-side institutions routed their execution through prime brokers, who held the information advantage and extracted spread on every transaction. Then hedge funds and quant trading firms used new technology to build direct access, co-location, and proprietary execution logic that went around the brokers entirely.
Did that bypass put the prime brokers out of business?
ADAM HEIMLICH: No — it stratified the market rather than eliminating anyone. Below a certain threshold of sophistication and scale, firms still use brokers. Above it, the relationship irreversibly rebalanced toward the buyer. Once a firm had built the capability to route independently, the cost of going back to broker dependence was higher than the cost of staying independent. The path around the broker, once established, becomes permanent. Advertising is at the equivalent inflection now. The pooled stack is the prime broker — it holds the model advantage, extracts margin, and owns the supply relationship. The largest advertisers have the volume to bypass it. What they lacked, until recently, was the capability.
For most of the last decade, “build your own bidder” was a slogan more than a practice. What changed?
ADAM HEIMLICH: It now has a working, public proof point. Hyundai tested running its CTV bidding through a containerized model built by Chalice and deployed inside the SSP OpenX. Instead of waiting for a platform to act on its behalf, Hyundai’s model bid via a direct, low-latency connection to the source. The pilot covered three vehicle models. Hyundai is now extending it across the entire fleet.
Hyundai’s CMO, Sean Gilpin, is quoted saying “we want to own that.” What’s the “that”?
ADAM HEIMLICH: The decision-making knowledge about which impressions and audiences turn into car buyers, at what cost. Gilpin is explicit that the goal isn’t the usual chase for a cheaper CPM — it’s reaching the right potential buyers and owning the apparatus that identifies them. He doubts a brand “gets a unique competitive advantage with off-the-rack platform elements” like Google’s Performance Max. That one sentence — “we want to own that” — is the entire business case for the enterprise architecture.
Why doesn’t the platform’s much larger dataset just win here? Breadth should beat one advertiser’s data.
ADAM HEIMLICH: The industry has learned to mistake breadth for predictive power, but breadth isn’t what the enterprise customer demands. Platforms optimize toward generic outcomes on signals shared across every advertiser they serve — they can’t optimize on Hyundai’s private definition of an auction worth winning. If Hyundai is trying to steal share from Toyota, a value prediction shared by Hyundai and Toyota is never going to work. More data about everyone can mean less data about the one thing a single advertiser is actually trying to predict.
You also talk about “the spread” — a $10 impression clearing at $4. Why does per-advertiser modeling change who keeps that $6?
ADAM HEIMLICH: Custom valuation finds better value, but the larger prize is the spread. In open RTB, the incentive has always been to take it clandestinely — every programmatic controversy from US v. Google to the fights over supply-path optimization is rooted in intermediaries positioned to capture spread without the buyer noticing. Per-advertiser modeling drags that contest into the open. The brand runs its own value function, bids into supply it reaches directly, and keeps the spread its own model finds. Surplus accrues to the buyer whose model found it — that’s a cleaner, more efficient market than one where intermediaries skim it in the dark.
Won’t platforms eventually learn to price the spread away once they see what these models are doing?
ADAM HEIMLICH: Two things stop them. First, when Hyundai builds and runs its own model, it knows things about that model the platform doesn’t — it can withhold data that would let a platform reconstruct its logic. A sell-side system can’t accurately price against a valuation computed from data it doesn’t hold. The pooled stack extracts spread precisely because it sees both the prediction and the clearing; split the two and the informational advantage runs the other way. Second, Hyundai is benefiting from competition among venues. Per-advertiser modeling is standardized by the IAB Tech Lab as ARTF, the Agentic Real-Time Bidding Framework. Index Exchange and Chalice co-developed the standard, OpenX adopted it second, and others are moving the same direction. An advertiser can deploy the identical model across every ARTF environment and compare what each impression actually costs. A venue that prices the spread away loses that bid to one that doesn’t. Portability turns the buyer’s model into a check on every venue at once.
If I’m a CMO with a nine-figure budget reading this, what’s the actual decision in front of me?
ADAM HEIMLICH: The enterprise alternative to pooled-stack architecture is a disaggregation play, and it isn’t theoretical. It has a customer base in the top 2,000 advertisers, who control the majority of open-internet spend and are increasingly aware that their private data and valuations are competitive assets no counterparty should be allowed to claim and use to train its own AI. It has working infrastructure, now that buy-side decisioning can be hosted under a standardized, per-advertiser framework. It has a proof point in Hyundai. And it has M&A catalysts of its own: the largest media buyer in the world paid a premium to hedge against the split, and the largest retailer in America paid $1.4 billion for an SMB platform to consolidate the network side.
Last one — in the institutional-trading migration, each firm had to build its bypass internally at great cost. Is advertising’s version cheaper?
ADAM HEIMLICH: It can be. Advertising’s version can be external and shared at the infrastructure layer without pooling any single advertiser’s private logic. Each model stays specific to one buyer’s objectives, one buyer’s data, one buyer’s competitive context — it can’t be commoditized or pooled without destroying what makes it valuable. Per-advertiser decisioning is per-advertiser by definition. Both architectures are now funded, public, and accelerating. The only decision left to a brand is which one it belongs in. For the advertisers with the most at stake, the finance precedent isn’t a forecast. It’s a description of what’s already happening.
To read Adam’s piece in its entirety, visit https://www.adamlovesadtech.com/the-two-architectures/