Let's get something straight.

If you've been buying CTV programmatically and congratulating yourself on your "convergent TV strategy," I have some news for you. You've been shopping at the gift shop and calling it the museum.

The programmatic CTV boom of the last several years has produced something remarkable: an entire industry of marketers who believe they are buying television, when in fact they are buying a carefully curated slice of television's leftovers. Premium-ish. Targetable. Measurable. And wrong.

This is the first in a three-part series about how CTV is breaking up with the programmatic ecosystem. And why that breakup is going to be very, very expensive for a lot of very comfortable middlemen who have been clipping coupons on other people's media budgets for years.

I reached out to executives across the industry for comment. Several responded. Their answers were more candid than anything you will read in a press release. We will get to those.

Buckle up.

The U.S. TV advertising market is roughly $90 billion a year.

Say that out loud. Ninety. Billion. Dollars. This is not a niche channel. This is not "emerging media." This is the most powerful advertising medium human civilization has ever produced, and it is absolutely enormous.

Of that $90 billion, streaming accounts for about $30 billion. Linear -- you know, the thing everyone keeps declaring dead at every single conference panel since 2015 -- still commands roughly $60 billion. The death of linear TV has been, to put it gently, a collective hallucination. Linear is not dead. Linear is sitting in the corner counting its money while the entire industry writes its obituary.

Fine. Focus on the $30 billion streaming number. That's where all the excitement lives, all the venture capital went, and all the breathless trade press coverage happens.

Take that $30 billion and cut it in half.

About $15 billion flows through programmatic pipes. The biddable, auctioned, DSP-accessible inventory that the ad tech world has spent a decade and probably $50 billion in venture capital building infrastructure to serve. This is what shows up in your campaigns. This is what your trading desk optimizes. This is what your SSP packages and your DSP buys and what four or five intermediaries take their cut of before a single dollar reaches a publisher.

The other $15 billion? That's the good stuff.

And here is where it gets specific. Ben Tatta, Chief Commercial Officer of Operative …one of the most widely used ad revenue platforms in the media business, the people who actually see how publishers transact -- put a precise number on it when I asked him directly.

"Although well over half of CTV inventory is sold via programmatic channels, I estimate that only 20 to 30% of premium CTV inventory is currently sold via programmatic channels."

Tatta defines premium the way the industry actually uses the word. Fifteen and thirty second spots. Fixed ad breaks. Commercially produced programming. Disney. Fox. The shows people actually watch and then argue about online. Not the long tail. Not the FAST channel filler. The real stuff.

His read on where that premium inventory actually lives is equally blunt. "The majority of premium CTV inventory is being sold directly on an a-la-carte basis or as part of a linear bundle." And increasingly through self-serve platforms like Disney Ad Manager and Roku Ad Manager that bypass the programmatic stack entirely while still providing automation.

Read that again slowly. The majority of premium CTV inventory. Not accessible programmatically. From someone whose company sits inside the actual transaction infrastructure of major publishers.

You cannot log into a DSP and buy the premier episode of a prestige drama. You cannot programmatically guarantee yourself a spot in the Super Bowl. You cannot auction your way into Olympics overtime. That inventory exists. It is for sale. It is just not for sale to you through the pipes you have been using.

As Andy Schonfeld has written plainly, programmatic CTV "represents only a portion of the total TV inventory available to advertisers." A portion. Not a majority. Not a representative sample. A portion that skews, structurally, toward what publishers could not sell any other way.

Here is the core problem stated as simply as possible.

The programmatic ecosystem built a very impressive looking map of CTV.

The map has targeting parameters and bid floors and supply path optimization and audience segments and frequency caps and seventeen acronyms your media agency uses to justify their retainer. It feels comprehensive. It feels sophisticated. It has dashboards. It has so many dashboards.

But the map is not the territory.

The territory is the full $90 billion TV market. The territory includes every impression that airs during a live sporting event, every slot in a prestige series, every piece of inventory that a major streaming publisher holds back for direct relationships because that is where the real money and the real partnerships actually live. The programmatic map shows you a piece of that territory. A real piece. A functional piece. But a piece.

The entire industry has been staring at the map and assuming it represents everything worth buying.

It doesn't. And the gap between the map and the territory is where billions of dollars in premium inventory have been hiding in plain sight while your DSP told you it was running at full optimization.

Here is where I want to be completely direct about what is actually at stake, because the trade press has been remarkably gentle about this and someone needs to say it out loud at full volume.

The programmatic ecosystem is not an abstraction. It is a collection of companies.

DSPs. SSPs. DMPs. Measurement vendors. Verification layers. Data brokers. Supply path optimizers. Curators. Resellers. And assorted other entities that exist for one reason: because media transactions flow through them. Every hop in that supply chain extracts a fee.

How much? I asked Tatta directly. His answer was more precise than the headline number that typically gets thrown around at conferences: "The fees tend to fluctuate based on the CPM value. A more accurate estimate would be fees ranging between 10% and 50% of the CPM rate." That range covers not just tech fees but data, targeting, ad verification, and everything else that accumulates as a dollar moves through the stack.

Ten to fifty percent. On a $15 billion biddable CTV market, even the low end of that range represents an almost incomprehensible amount of money flowing to infrastructure that adds zero audience value.

Now imagine that a meaningful portion of that $15 billion -- plus the other $15 billion that never went programmatic in the first place -- starts moving through direct, automated pipes that bypass the ecosystem entirely. No DSP. No SSP. No verification tax. No reseller margin. No fees on fees on fees on fees.

The intermediaries don't just take a smaller cut. They take no cut. They are not in the transaction. They do not exist in that transaction.

This is not a rounding error. This is not pricing pressure. This is a structural threat to the business model of a significant portion of the ad tech industry. I am genuinely baffled that more people are not saying this plainly in public instead of whispering it in the hallways at every conference where the DSP reps are buying the drinks.

I asked Tatta whether publisher motivation to support direct pipes is a temporary negotiating posture or something more fundamental.

He did not hedge. "I believe this is a permanent shift among large publishers with premium inventory."

He drew a distinction that gets consistently lost in the broader disruption narrative. For mid-to-small publishers with less premium inventory, aggregation of demand through DSPs and SSPs remains genuinely necessary. The pipes still serve a real purpose at the long tail. But the structural threat is concentrated at the top of the market -- exactly where the most valuable inventory and the highest CPMs live. That is not a coincidence. That is the entire point.

And then Tatta said something that should be printed, framed, and hung on the wall of every DSP product team in the industry. When asked what happens to DSPs and SSPs as direct automation scales, he pointed to a development the trade press has barely begun to process: "Especially with the emergence of agent-to-agent transactions, it seems as though large premium publishers will follow a more direct model."

Agent-to-agent. AI systems on the buy side and the sell side conducting transactions based on explicit, transparent rules. No black box. No auction mystery. No opaque fee extraction at every hop. If you think direct automation is the disruption, agent-to-agent is the disruption after the disruption. And the companies building the current direct infrastructure are the ones best positioned to build the next layer.

Pay attention to that sentence. It is the most important one in this piece.

How did an entire industry convince itself that programmatic was a sufficient CTV strategy? How did we spend a decade building auction infrastructure for a medium that was never really an auction medium?

Partly it's the seductive familiarity of digital.

Programmatic display worked. Programmatic search worked. Social worked. Everything in digital moved toward auction-based impression-level buying and it made sense because supply was vast, fragmented, and effectively infinite. You needed a machine to navigate it. Without a DSP you were drowning.

CTV is not that. CTV is the structural opposite of that.

The supply is concentrated. Profoundly concentrated. The overwhelming majority of what anyone actually watches on streaming flows through a remarkably small number of publishers. Ask someone what they watched last night. The answer almost always references one of maybe ten services. That is it. Ten companies. The entire premium CTV market, for all practical purposes, lives in ten places.

When supply is that concentrated, you do not need an auction to find it. You don't need a DSP to navigate a fragmented landscape because the landscape is not fragmented. The challenge is not discovery. It is access, relationship, and execution. And programmatic was not built for that problem. It was built for the exact opposite problem.

The industry took the tools of a fragmented display world, dropped them into a concentrated TV world, and then acted genuinely surprised when the fit was terrible. And then built more tools on top of the terrible fit. And then charged fees for those tools. And then hired salespeople to explain why the fees were reasonable.

It is actually kind of impressive when you think about it.

Before the DSP executives reading this compose an angry reply, let me be precise about what is and is not being argued here.

Tatta put the future of programmatic in terms that are more interesting than either the cheerleaders or the critics have managed. He started with a point the industry has needed someone to make for years: "Unfortunately the term programmatic has been bastardized. In a pure sense, the term programmatic is about automation, self-service, self-governance. Today, programmatic is defined as disintermediation, fraud, ad tax, gray market."

That is a devastating sentence from someone who has watched this market from the inside for years. The word that was supposed to mean efficiency has become synonymous with everything that makes buyers distrust the ecosystem they are dependent on.

His view of where it goes from here is genuinely constructive: "Over time I believe the programmatic ecosystem becomes more like AI middleware, where buy-side and sell-side agents conduct transactions based on explicit, transparent rules, versus everything running through a black box. Given the state of programmatic today we might have to call it something else -- but it basically evolves to a more pure form of programmatic with fewer intermediaries."

Fewer intermediaries. Explicit rules. Transparent transactions. Buy-side and sell-side agents dealing directly with each other. That is not the programmatic ecosystem as currently constituted. That is a fundamentally different thing wearing the same name.

The companies that survive the transition are the ones that understand that distinction and move toward the new model before the market forces them to. The ones that don't will spend the next five years defending infrastructure that the market has already decided to route around.

History is not kind to toll booths when someone builds a bypass.

Part two goes inside the fee stack, the fraud problem, and the brand safety fog that nobody in the programmatic ecosystem has any real incentive to clean up. It is for the people who already know something is wrong and want the specific numbers.

Part three is about what the alternative infrastructure actually looks like, who is building it, and what it means for every player in the current ecosystem when the parallel operating system finishes being built.

But here is the headline, stated as plainly as possible:

CTV is leaving the programmatic ecosystem. Not entirely, not overnight, and not without a fight from the companies whose revenue depends on staying in the middle. Programmatic retains a real role for retargeting, certain audience use cases, and remnant inventory monetization. It does not disappear.

But the gravity is shifting. The premium inventory is moving. The publishers want it to move. The buyers are learning to want it to move. And the technology to make it happen without the manual friction that used to make direct buying inaccessible -- that technology exists right now.

The companies that built their businesses on being between the buyer and the publisher?

They should be very, very focused on what comes next.

Because the map just got a lot smaller. And the territory is not waiting around.

Next: Part 2 -- The Tax, The Fraud, and The Fog. What programmatic CTV is actually costing you, and why the people who know won't say it out loud.

Upgrade Now