For the first time, US advertisers committed more upfront money to connected TV than to primetime linear television. That is a genuine milestone, because the upfront exists because of primetime. It is also the most likely fact in marketing right now to be misread as a buying instruction.

What actually happened

EMARKETER forecasts that US connected TV upfront ad spending reaches $17.73 billion in 2026, against $16.98 billion for primetime linear TV. A gap of roughly $750 million on a market worth almost $35 billion combined is not a landslide. It is a crossover, and crossovers only happen once.

The upfront matters here for a reason that is easy to miss. It is the annual market where advertisers commit budget months before a single impression is served, and it was invented for primetime network television. Streaming winning that specific daypart, in that specific market, is the industry conceding where the audience finished moving. EMARKETER's own wording is that CTV is poised to "surpass linear TV in prime-time upfront ad spending for the first time".

The wider number is bigger and less symbolic. Total US CTV ad spend is forecast at roughly $37.95 billion in 2026, growing around 15 percent year over year. Total advertising is growing somewhere in the 6 to 8 percent range. Linear is shrinking. So the money is not being created, it is being moved.

The part that gets skipped: an upfront commitment is a reach purchase made a year in advance by companies with brand budgets. Nothing about that milestone describes the economics facing an advertiser who is spending five figures a month and measuring cost per lead every Monday.

What streaming attention costs

The honest way to evaluate any channel shift is to put the prices next to each other on the same unit. Here is what a thousand impressions costs across the places most advertisers can actually buy today.

Channel Average CPM What you are buying
Google Display Network$3.12Cheap, skippable, low attention
Programmatic open exchange$5.85Remnant inventory, quality varies
Private marketplace$8.20Vetted publishers, negotiated
Meta, blended$14 to $15Targetable, measurable, crowded
FAST channels$15 to $25The cheap end of streaming
Connected TV, blendedabout $26Full screen, unskippable, high completion
Premium streaming, programmatic$25 to $45Named services, real shows
Premium direct and live sports$45 to $65Appointment viewing, scarce
Addressable, first party targeted$45 to $85Your list, on a television screen

Read the top and the bottom of that table together. Streaming costs roughly eight times display and close to double Meta. That is the price of a full screen with the sound on and no skip button. It can absolutely be worth it. It is never worth it by default, and no upfront milestone changes that arithmetic.

If you want the same treatment applied to search and social specifically, the numbers I keep updated live on the Paid Media Cost Index.

The two curves nobody put on the same chart

Here is the finding that got almost no coverage, and it is the reason this story is worth your time rather than just your attention.

Streaming is getting cheaper. Amazon switched ads on for Prime Video in early 2024 and dumped an enormous amount of new inventory into the market at once. Its CPM fell from around $35 to roughly $28 within a year. That is what supply does to price, and more supply keeps arriving as every service launches an ad tier.

The channels you already live in are getting more expensive. Google Display Network CPMs rose about 7 percent year over year, described in the benchmark data as the steepest CPM inflation since 2021. Private marketplace inventory is forecast to climb another 10 to 14 percent, and the gap between open exchange and PMP pricing has widened more than 60 percent since 2024.

So you have one curve falling and one curve rising, and every trade headline is about the crossover in the upfront rather than the convergence in price. The convergence is the thing that will eventually change what a normal advertiser buys. It has not changed it yet. Roughly $26 against roughly $14 is still close to double, and the cheaper option is the one where you can see which creative drove which conversion.

What to watch instead of the upfront: the spread between blended CTV CPM and blended Meta CPM. When that spread closes to something like 30 percent, the conversation stops being about brand budgets and starts being about media plans. Track it once a quarter. It is a far more useful number than any milestone.

Who this changes something for

Four situations, and only two of them involve doing anything.

You spend under roughly $15,000 a month on paid media. Nothing changes. Video reach buys need frequency across a large enough audience to register, and at a $26 CPM a small budget buys a number of impressions too thin to move anything. Spend it where intent already exists.

Your search program is capped by demand, not by budget. This is the one case where the news is actionable. If you are already capturing most of the searches that exist for what you sell, more budget in search buys you worse traffic. The constraint has become how many people know the category exists, and that is exactly the problem video reach solves. Before you conclude you are here, check it properly rather than assuming, which is most of what a paid media audit is for.

You sell something considered, high ticket and geographically bounded. Addressable CTV against your own customer list, at $45 to $85 CPM, is defensible when a single closed deal is worth thousands. Run it as a named test with a holdout, not as an always on line item.

You still have lost impression share in search. Buy that first. Every time. A channel where someone has already typed what they want will outperform a channel where you interrupt them, until you have exhausted it. The full comparison of how those two behave is in Google Ads versus Facebook Ads.

What I would do with the next budget cycle

In order, and the order is the whole point.

1. Pull lost impression share due to budget on your search campaigns. If it is above 10 percent, you have unbought demand and this entire article is background reading. Fund that first, because it is the cheapest growth available to you and it needs no new creative, no new vendor and no new measurement approach.

2. If search is genuinely capped, treat CTV as a reach line and measure it like one. Geo holdout or a matched market test. Look at branded search volume, direct traffic and blended cost per acquisition over the test window. Do not accept a view through conversion window as evidence, and be careful with any vendor who offers you one as the headline metric.

3. Start at the FAST tier before the premium tier. At $15 to $25 you can learn whether video moves anything for your category at all, at less than half the cost of finding that out on premium inventory. If it works there, upgrade. If it does nothing there, premium was unlikely to save it.

4. Recheck the CPM spread in three months. Both curves are moving. A decision that is wrong today can be right by the next planning cycle, and the only way to know is to look at the number rather than the headlines.

The upfront crossover is real and it is worth knowing. It is a fact about how large brands allocate committed reach budgets a year in advance. It is not a fact about your cost per lead, and anyone presenting it as one is selling inventory.

Sources

Forecast figures are forecasts. EMARKETER revises them, and the CPM benchmarks above are market averages that will differ from what your account actually pays. Treat every number here as a starting point for your own measurement, not a substitute for it.

Questions

Is connected TV actually getting cheaper?
At the blended level, yes. Every streaming service launching an ad supported tier adds inventory faster than demand is arriving, and Amazon Prime Video is the clearest example, falling from around $35 to roughly $28 within a year of switching ads on. Premium and live sports inventory is the exception, because that supply is genuinely scarce and its price is holding.
Does this mean I should move budget out of Google Ads?
Almost certainly not. Search captures demand that already exists and streaming creates awareness that does not. They are not substitutes. The only case where moving budget makes sense is when your search campaigns have run out of demand to buy, which you can verify by checking lost impression share rather than guessing.
What is a realistic minimum budget for a CTV test?
Programmatic platforms will accept far less, but at a $26 CPM a meaningful test needs enough frequency across a defined geography to be detectable at all. In practice that means a dedicated test budget on top of your existing paid media, not carved out of it, and a business already spending comfortably into five figures a month. Below that, the test cannot tell you anything reliable.
How do I measure connected TV without last click attribution?
Geo holdout is the practical answer. Run the campaign in a set of markets, hold back a matched set, and compare branded search volume, direct sessions and blended cost per acquisition across both. It is less precise than a click path and far more honest than a view through window, which will credit streaming for conversions it had no part in.
Why did the upfront crossover happen now?
Because the audience finished moving and the inventory finished arriving. Ad supported tiers on the largest services turned streaming into something a media buyer can commit to a year ahead at scale, which is what an upfront requires. Linear primetime did not collapse so much as it stopped being the largest place to buy committed reach.

This is a news piece, which means it is about something that changed. If you want the durable version of how these channels compare and when each one earns its budget, that lives in the Diwizi blog. More recent developments are on the news page.