Your model is oversimplified in a way that downplays the value of ads to the manufacturer. They don't reach out to sponsors and get a $50 static offer per TV in

Your model is oversimplified in a way that downplays the value of ads to the manufacturer. They don't reach out to sponsors and get a $50 static offer per TV in deals. They do some math and figure out that they can make at least $X per customer on average over the lifetime of the TV by selling ad slots dynamically. The subtle difference here is that because the sponsorships can be updated live across TVs that have already been sold, the actual value of each TV sale can be made to go up after the date of purchase by updating software and/or changing the ad deals. So the manufacturer isn't pricing the TVs at a discount precisely equal to the ad revenue they receive per TV, they have to price the TVs based on a complicated formula that includes both a rough estimate of the minimum value of ad deals and customer willingness to pay (keeping in mind that customers are choosing their willingness to pay based on a landscape that has no ad-free models!). And what's more, the manufacturer is free to alter the deal after the sale is made to try to make a larger profit per-TV than was originally priced in. You make it sound like it's a reasonable outcome of an efficient market, but the current situation—where one party can and does alter the deal retroactively and unilaterally—does not create an efficient market!