Streaming Hits 49% of U.S. TV Viewing; CTV Ad Spend to Top $38B in 2026

Streaming Crosses Half of U.S. TV Viewing

Nielsen's July 2026 Gauge put streaming at 49% of U.S. TV time, with YouTube alone at a record 14.2% share. EMARKETER projects CTV (connected television) ad spend at $37.95 billion for 2026, up 14.5% year-over-year, and for the first time, upfront commitments to CTV will exceed primetime linear.

The surface story is streaming won. The operational question is who gets paid differently now.

The money is moving, but not uniformly

YouTube's 14.2% share is the single largest slice of U.S. TV time, and it clears at CPMs well below the floors Disney or NBCUniversal set for premium streaming inventory. That gap matters. Total CTV dollars are growing, but the fastest-growing chunk is also the cheapest. Blended CPMs fall even as total spend rises. So the $38 billion headline overstates the yield story.

Roku, Netflix, and Disney Advertising all benefit from the narrative forcing CFOs to reallocate. Broadcast groups like Sinclair, Gray, and Tegna lose their last argument for linear-first upfront inclusion. Over two years, holding companies will collapse separate linear trading desks into unified video teams, reducing the number of DSP relationships they maintain. The volume flows to The Trade Desk (the largest independent ad-buying platform) and Magnite (the leading independent sell-side platform). The mid-tier SSP layer consolidates.

The cost nobody is counting

More CTV volume routed through more measurement vendors is not automatically better unit economics than the linear book it replaced. Buyers now pay Nielsen, possibly iSpot or VideoAmp, possibly a clean room, to answer a reach question that linear answered with one currency. That measurement tax does not lift reach. It just lets you trust the reach you bought. For publishers, the question is whether the incremental CTV dollar clears enough margin after identity, measurement, and SSP fees to beat the linear dollar it replaced. For YouTube-grade inventory, often no.

The workflow problem is just as real. Linear sells guaranteed audience points. CTV sells impressions that deliver dynamically. Frequency capping breaks first: a viewer watching the same show across devices gets hit with the same ad repeatedly because the two systems do not share identifiers. Make-goods break second: when a campaign underdelivers, the reconciliation math works differently in each environment, and resolving it by hand eats Q3 and Q4 capacity.

The measurement question decides everything

Whether this crossover is a genuine repricing of CTV inventory or a nominal dollar win against a collapsing linear base comes down to one thing: whether cross-platform measurement gets credible fast enough to collapse the buyer's risk discount. If Nielsen ONE and its challengers converge on a currency the buy side trusts, premium CTV CPMs reprice upward and the volume story holds. If measurement stays fragmented, every CPM carries a doubt discount into 2027.

The timing argument favors fragmentation. Measurement standards do not consolidate on an upfront's schedule.

Our call: U.S. CTV ad spend will grow faster than average CTV revenue per impression in 2026. YouTube's record share drags the blended CPM down while total dollars climb, and measurement fragmentation gives buyers a reason to demand a discount on reach they cannot fully deduplicate. We revisit when EMARKETER reports full-year 2026 figures by March 2027. We're wrong if average CTV CPM grows as fast as or faster than total CTV spend.

Streaming just crossed half of U.S. TV viewing. Nielsen's July 2026 Gauge put streaming at 49%, with YouTube alone at a record 14.2% share. EMARKETER says CTV ad spend hits $37.95 billion in 2026, up 14.5%, and for the first time, advertisers will commit more upfront dollars to CTV than to primetime linear.

Here's the frame. The surface story is "streaming won." The decision for an ad-tech operator is different: when the upfront money formally crosses over, whose rate cards, trading desks, and measurement contracts get rewritten, and who gets squeezed. This is hard to undo. Once a holding company CFO reallocates a linear budget line into a video investment team, that budget does not march back to broadcast next year. The deadline is the 2027 upfront. That negotiation sets the new baseline, and nobody wants to anchor low.

---

The Market Analyst

The $38 billion headline is a nominal-dollar win against a shrinking linear base, so don't read it as yield parity. The interesting number is YouTube's 14.2%. It is the single biggest slice of TV time in the country, and it does not clear at Disney or NBCUniversal CPM floors. That gap splits the winners. Roku, Netflix, and Disney Advertising all benefit from a narrative that forces CFOs to move money. The losers are the broadcast groups, Sinclair, Gray, Tegna, who just lost their last argument for linear-first upfront inclusion. The structural play over two years: holdcos collapse separate linear trading desks into unified video teams, which thins the number of DSP relationships they keep. Volume flows to The Trade Desk and Magnite. The mid-tier SSP layer gets consolidated.

In plain terms: the money is moving to streaming, but it is not all the same quality of money, and the firms that sell measurement and identity collect the toll.

The Skeptic

Steelman the bull case and it cracks. "CTV upfront surpasses linear" compares a growing number against a collapsing one. That is not evidence that a CTV impression yields what a linear GRP yielded. The CPM gap between premium streaming and linear has not closed in spots, it has widened. And buyers are committing these dollars against metrics they do not fully trust. Nielsen ONE adoption is still incomplete, iSpot and VideoAmp are still fighting for legitimacy, and every one of those cross-platform currency fights means the buyer is guessing at duplicated reach. That uncertainty is a cost. It sits inside every CPM as a discount the buyer demands for committing blind. Convergence of eyeballs is real. Convergence of dollars-per-eyeball is a story nobody has the measurement to prove.

For the non-specialist: advertisers are putting big money down on audiences they cannot yet count the same way across screens.

The Operator

The crossover lands on planning desks Tuesday morning as a workflow collision. Linear sells guaranteed audience points. CTV sells impressions that deliver dynamically. Yield and ad-ops teams at the hybrid shops, Peacock, Tubi, the Roku Channel, have to run both books at once. The first thing that breaks is frequency capping across environments. A viewer watching the same show on an app and a connected TV gets hammered with the same spot because the two systems do not talk. The second break is the make-good. When a campaign underdelivers, the math for what you owe back works one way in linear and another way in streaming, and reconciling them by hand eats Q3 and Q4. Teams will paper over this longer than they should, because the alternative is re-platforming mid-cycle.

Said simply: the money moved faster than the plumbing, and the plumbing is where next year's margin leaks.

The CFO

Everyone is celebrating the $38 billion in, nobody is costing out the profit per impression on the way through the supply chain. More volume at a lower effective CPM, routed through more measurement vendors each taking a cut, is not obviously better unit economics than the old linear book. The real cost here is the measurement tax. Buyers now pay Nielsen, maybe iSpot or VideoAmp, maybe a clean room, to answer a question linear answered with one currency. That is new spend that does not lift reach, it just lets you trust the reach you bought. The payback question for a publisher: does the incremental CTV dollar clear enough margin after identity, measurement, and SSP fees to beat the linear dollar it replaced? For YouTube-grade inventory, often no.

Plainly: more revenue is coming in, but the cost of proving what you sold is rising alongside it.

---

Where they split

Two real disagreements. The Market Analyst sees a volume win that rewards scale players and consolidates the middle. The Skeptic says the win is nominal and the measurement doubt is a permanent discount, not a transition cost. They cannot both be right about whether the money that crossed over is high-quality money.

Second: the Operator and the CFO disagree with the headline on where value leaks. The headline treats the crossover as arrival. Both of them treat it as the start of a cost problem, the Operator in broken workflows, the CFO in a measurement tax that eats the incremental margin.

What it hinges on

Whether cross-platform measurement gets trustworthy fast enough to collapse the buyer's risk discount. If Nielsen ONE and the challengers converge on a currency the buy side actually believes, the Market Analyst wins and CTV dollars reprice up. If they stay fragmented, the Skeptic wins and every CPM carries a doubt discount into 2027. The council leans toward the Skeptic on timing. Measurement standards do not consolidate on an upfront's schedule, and the YouTube share drags the blended CPM down regardless. Before committing 2027 budget, a buyer should pressure-test one thing: what deduplicated reach am I actually getting across YouTube, premium CTV, and the residual linear, and who do I trust to tell me.

---

Prediction: U.S. CTV ad spend will grow faster than CTV ad revenue per impression in 2026, confirmed when EMARKETER reports full-year 2026 CTV spend and average CTV CPM figures by March 2027.

Confidence: Medium — YouTube's record 14.2% share drags blended CPM down while total dollars climb.

Why: The $37.95 billion headline is 14.5% dollar growth, but the fastest-growing slice of CTV time is YouTube at a record 14.2% share, and YouTube inventory clears well below premium walled-garden CTV floors from Disney or NBCUniversal. When the cheapest high-volume inventory grows fastest, average revenue per impression falls even as total spend rises. On top of that, measurement fragmentation means buyers demand a discount for committing against reach they cannot fully deduplicate, which pushes effective CPMs down further. The opposite outcome, CPMs rising alongside volume, would require premium inventory to grow faster than YouTube and the measurement doubt to resolve, and neither is happening on this timeline.

Revisit by 2027-03-31: We're right if EMARKETER's full-year 2026 figures show CTV ad-spend growth outpacing average CTV CPM growth. We're wrong if average CTV CPM grows at or faster than total CTV ad spend.