The Cheapest Truck Wins

Thirteen NBA teams and seven NHL teams woke up this spring as local-television free agents.

When Main Street Sports Group announced in April that it would wind down — the corporate successor to Diamond Sports, the operator of the FanDuel Sports Network portfolio, the last big load-bearing beam of the regional sports network era — twenty franchises lost their local broadcast home in a single stroke. The bids they have been fielding since are not rescue offers. They are haircuts. DAZN has been courting the orphaned teams with rights-fee cuts of as much as 40 percent. Per SBJ’s reporting on the scramble, Fubo has been offering minimum guarantees in the range of $8–15 million per team; Gray, Scripps, and Nexstar have been coming in below $10 million. And then there is the line in Yahoo Sports’ write-up of that reporting that I cannot pretend to read neutrally: “Even the free streaming service Victory+ is making a play for these rights.”

I should stop here and say plainly: I am the Chief Strategy Officer of Victory+. Everything about the company in this essay comes from third-party reporting — SBJ and the trade press — not from inside the building, and nothing here should be read as the company announcing anything.

The coverage of this repricing has treated it, almost uniformly, as a story about rights fees. Who pays $12 million, who pays $8 million, who demands a 40 percent cut, which league centralizes and which one doesn’t. I think that framing misses the actual contest. When every bid on the table is a haircut, the rights line stops being where the war is won. The war moves to the cost side — specifically, to the question of who can produce seventy or eighty professional-grade game telecasts a year for the least money. And the cost of producing a game telecast is, right now, being rewritten by a wave of AI production tooling that almost nobody covering these deals is pricing in.

The cheapest truck wins. That is the thesis. Let me earn it.

The Vacuum

The shape of the vacuum matters, because it determines how long the window stays open.

Main Street’s wind-down, telegraphed by SBJ in January and executed at the end of the NBA and NHL seasons in April, is the terminal event of a collapse I documented at length in my Q2 research report on local sports rights. The one-paragraph version: the RSN model ran on cable-bundle subscriber fees; the bundle collapsed; a $9.6 billion leveraged acquisition became the largest concentrated repricing event in the history of American sports media. MLB got out early — the league’s clubs terminated en masse in February, and MLB Local Media now produces local telecasts for 14 of the 30 clubs, with the Angels buying their RSN outright and selling Angels.TV at $74.99 a season, the Braves launching BravesVision, and the commissioner’s office openly targeting all 30 clubs’ local rights in one national in-market package by the end of the 2028 season, per CNBC.

Baseball, in other words, answered the vacuum by absorbing the production function into the league. The NBA and NHL teams stranded this April have no such apparatus waiting. What they have is a menu of discounted bridge offers — and a league office, in the NBA’s case, actively assembling a centralized in-market streaming package targeted at the 2027-28 season, and urging its stranded teams toward one-year deals in the meantime. YouTube TV, per Awful Announcing and SBJ, is interested in the local package only if twenty or more teams come bundled together. The message under the message is that the current scramble is an interregnum. Whatever deals get signed this summer are placeholders until the map gets redrawn.

Some of the placeholders are revealing. Scripps signed the Pistons in May — its first NBA team, reported in the $8–10 million range, with a DTC app promised and details to follow. Gray extended the Suns and Mercury for four years over the air — a long-term re-up of what had been the first modern NBA deal to leave an RSN for broadcast television. DAZN, having walked away from rescuing Main Street itself, announced in April that it would merge ViewLift into its own operation — absorbing, in one roughly $100 million transaction, the streaming plumbing behind NESN, Monumental, Altitude, Space City, and the Chicago Sports Network.

A one-year deal at a 40 percent discount is not a partnership. It is a holding pattern. Everyone in this market — the broadcast groups, the streamers, the league offices — is buying position ahead of the real repricing in 2027-28. Which is exactly why the cost side matters so much right now. The operator who can run profitably during the holding pattern, at haircut rights fees, on one-year terms, with no guarantee of renewal, is the operator who gets to be standing there holding relationships and infrastructure when the big package finally gets assembled.

The Truck

To see why production cost is the live variable, you have to look at what the RSN model actually paid for.

The regional sports network was funded by a per-subscriber fee — $5 to $8 a month, every month, from every cable household in the team’s territory, whether that household watched a single game or not. I walked through the mechanics of this in the Q2 report; the short version is that the marginal fan, the household that paid the fee without watching, was the entire pricing structure. And that structure funded a production standard built to national-broadcast expectations: the truck outside the arena, the camera positions, the full crew, the announce team, the replay and graphics operators, the pre-game and post-game shows wrapped around 70-plus telecasts a year. I am not going to pretend there is a clean public number for what an RSN-grade telecast costs per game — operators guard those numbers, and the setups vary enough that a single benchmark would mislead. What is public, and structural, is this: the production line was priced for a world in which millions of non-viewers paid for it.

That world is gone. The cost line it funded is still here.

This is the inheritance problem at the center of every deal being negotiated this summer. A team leaving the RSN system does not just lose a rights fee; it inherits, one way or another, a production obligation that was scaled to someone else’s business model. Either the team pays for production itself, or the league absorbs it (MLB’s answer), or the new rights partner takes it on inside an $8–15 million all-in number that has to cover rights and trucks and people. Every one of those paths runs through the same question — what does it actually cost to put a professional game on a screen in 2026? — and the honest answer is that the number is falling fast, unevenly, and mostly out of view of the people writing about the deals.

I wrote in the spring about how the production-cost collapse rewired the economics of minor league broadcasting — how the mid-tier broadcast that used to require an RSN budget can now be produced at a fraction of historical cost, and how the decision-makers of 2020 cut teams whose economics had already flipped without anyone noticing. What has changed since I wrote that essay is the arrival of a specific new layer on the cost curve. The AI layer.

The New Stack

Here is where the production stack actually stands, tier by tier, as of this summer.

At the youth and lower tiers, fully automated production is not an experiment; it is the default. Pixellot’s automated camera systems, distributed through partners like NBC’s SportsEngine, were on track to capture 1.5 million games in 2025, per SBJ’s reporting on the company’s latest raise — a company-supplied count, but the order of magnitude is the point. Spiideo launched AI Highlights in May, built with the AWS Generative AI Innovation Center. FloSports has deployed what it calls an “AI Producer” on FloWrestling — a virtual camera system with no human operator at all. The pattern across all three is the same and worth stating precisely: fully automated AI production exists today only where there was never a crew to replace. Nobody at the youth tier lost a job to Pixellot, because the alternative to the robot camera was no camera.

But the tooling does not stay at the tier where it was proven. Earlier this year AWS launched automated reframing of live horizontal feeds into vertical video — and the early adopters are not youth soccer platforms. They are Fox Sports Digital and NBCUniversal. This is, to my eye, the clearest pure production-cost-compression product in the market right now: a task that used to require human editors cutting vertical versions of live content, done by software, at the tier of the biggest broadcasters in the country. It is not a demo. It is deployed, on premium feeds, by companies whose entire business is premium feeds.

The premium tier itself tells a parallel story, and an instructive one. The World Cup that just ended was produced — all 104 matches, roughly 9,000 hours of content — through about 60 servers running as a private cloud in Dallas, with signal processing done in software on commodity hardware, per SVG’s excellent reporting from the IBC. Almost none of that is AI, and the trade press’s habit of writing “AI” when it means “software” has muddied this story badly. Fox called the tournament the biggest production in its 32-year history, and the AI it did announce sat at the derivative edge — the AWS vertical reframing of its digital feeds — not in the telecast itself. The humans stayed in every creative chair. What dissolved was the hardware underneath them: the dedicated truck, the purpose-built signal chain, the physical infrastructure that made every additional telecast expensive. The truck became software. That is the precondition for everything else, because once production is software-defined, AI features slot into a pipeline instead of requiring a forklift.

And at the top of the automation curve sits ESPN’s SC For You — the AI-personalized SportsCenter, built on WSC Sports’ engine, processing something like 16,000 game feeds a year with 25 tagged studio shows and a staff of 162 writers and editors feeding the pipeline, per SBJ’s reporting last November. I cite it not because a local operator needs to replicate it, but because it demonstrates the shape of the mature system: not zero humans, but a small number of humans supervising an automated content line whose per-unit output cost would have been unthinkable five years ago.

So the stack, top to bottom: software-defined infrastructure proven at the World Cup tier, AI reframing deployed at the Fox/NBCU tier, AI highlights and automated cameras proven at the youth tier and climbing. The direction of travel is unambiguous. Automation enters where there was never a crew, gets hardened at volume, and moves up-market one tier at a time — and the tier it is arriving at next is exactly the tier the RSN collapse just vacated.

The Equation

Now put the two halves together, because the arithmetic is the essay.

A local rights deal at a 40 percent haircut is a catastrophe if your cost structure is the old one. The legacy operator — the broadcast group running RSN-grade trucks, the team production department staffed to 2019 expectations — cannot make $8 million a year work against a cost line built for a $5-per-subscriber world. This is why the incumbent bids all cluster at the same uncomfortable altitude: below $10 million, hedged, one year at a time. The bidders are not being cheap. They are being honest about what the old production model lets them afford.

But the same deal is a different proposition entirely for an operator whose production stack was born after the collapse. If your game telecast runs on software-defined infrastructure instead of a truck, if your highlights are cut by a model instead of a night shift, if your vertical and social derivatives are generated by an AWS service instead of an edit bay, then your cost per telecast is not the incumbent’s cost minus ten percent. It is a structurally different number. And at a structurally different number, a haircut rights fee stops being a haircut. It becomes a margin.

When the rights fee takes a 40 percent haircut, the war is not won on the rights line. It is won on the production line.

I want to be careful about what I am and am not claiming. I am not going to invent a per-game cost comparison; no operator has published one and the configurations vary too much for a single number to be honest. The claim is structural: the spread between old-stack and new-stack production cost is now wide enough to decide who can survive the terms this market is actually offering. Two bidders can look at the same $8 million-a-year, one-year, no-guarantee deal, and for one of them it is value-destroying and for the other it is a business.

There is a larger irony sitting on top of this, and it is my favorite fact in the whole landscape. Institutional capital has spent the spring pricing sports as AI insurance — Axios reported in May that sports deals are becoming “AI protection,” JPMorgan’s Mary Callahan Erdoes called sports “the antithesis of AI” in June, and PitchBook reported within the same five weeks, citing Arctos, that private equity is betting on live events as the thing AI cannot replace. The thesis is right as far as it goes: the game itself is the one media asset a model cannot synthesize. But nobody making that argument has priced its corollary. AI cannot replace the game — what it can replace is a large share of the cost of televising the game. Production-cost compression works entirely in favor of whoever holds the rights and entirely against whoever’s margin was the production. The same technology wave that makes the rights more valuable makes the legacy apparatus around the rights less necessary. Sports is AI insurance for the asset owner and AI exposure for the truck.

The Empty Shelf

There is a second half to this story, and it is the part I find genuinely strange.

I spent part of this summer looking for a single shipped AI product feature at the local and team level of American sports streaming. Not a press release, not a partnership announcement, not an “exploring opportunities” quote — a feature a fan can touch. The search covered the broadcast groups building local portfolios, the TEGNA-style hybrids, MLB’s league-operated local channels, the ViewLift-built platforms that DAZN just absorbed, and the team-branded DTC apps that launched into the vacuum. I found nothing. No AI personalization, no automated highlights surfaced to fans, no catch-up features, no conversational anything. I want to hold this finding at its correct weight — it is an absence of evidence from targeted searching, not proof of a negative — but the absence is broad, and it is consistent.

Meanwhile, at the national tier, the product shelf is filling up fast. Amazon ships Rapid Recap and Key Moments on NBA games. ESPN ships Catch Up to Live and a personalized SportsCenter. Peacock is beta-testing AI-driven vertical cropping of live NBA broadcasts. The national platforms are lapping the field on product while the entire local war is being fought on a single axis: price. Every bid in SBJ’s reporting — the 40 percent haircut, the $8–15 million, the sub-$10 million broadcast offers — is a price position. Not one operator in the local market is currently differentiated by what their product does.

I read that gap as the actual opportunity hiding inside the post-RSN scramble. Fighting on price is what markets do when no one has anything else to fight with. The first local platform that ships national-grade product — highlights that find the fan, catch-up for the viewer who tunes in down two goals in the second period, a local telecast that behaves like it knows who is watching — gets to compete on something other than the size of its haircut. In a market where every offer looks like every other offer minus a million dollars, being the one platform a fan actively prefers is worth more than any line item in the deal memo.

The cheapest truck wins the current round. The smartest software wins the next one. The rare operator positioned to hold both ends of that sentence at once is the one to watch.

Which Model Captures Which Markets

My Q2 research report ended on a deliberately open question. The RSN era is over — that part is settled — and four replacement models are competing for the freed rights: league-direct production, free over-the-air broadcast, team-owned DTC, and free ad-supported streaming. The question the report posed was which replacement model captures which markets, on what terms, and at what valuation when local rights next reprice.

This essay is my answer, or at least the first half of one. The model that captures the market is not a distribution model at all. Distribution is the visible layer — OTA versus DTC versus FAST versus league app — and the report is right that most franchises will end up running hybrid stacks of two or three. Underneath every one of those stacks sits the same invisible layer: a production line, priced either for the world that ended in April or for the one that started. The operators built on the old line will keep bidding what the old line lets them bid, and they will keep losing ground a million dollars at a time. The operators built on the new one can pay the same haircut prices, clear margin at them, survive the one-year holding pattern, and arrive at the 2027-28 repricing with working infrastructure, fan relationships, and a P&L that works — while everyone else arrives with a truck they can no longer afford to park.

Twenty teams became free agents this spring. Nobody bidding for them is going to win on the rights line, because there is nothing left to win there — every number is a haircut and everyone knows it. The deals will be announced on the rights line. They will be won on the production line.

The cheapest truck wins. Then the truck stops being a truck at all.