That gap is the question I get asked most, usually by somebody running a facility who has read the same recovery numbers I have and can’t find them anywhere in their own books. My argument on the panel was that you can’t explain this industry right now by looking at any single part of it. The answer I’ve landed on since is that we’re living through three transitions at the same time, which is genuinely different from previous industry transitions.

What we asked for, and got

Our August Screen Production Index has “smoothed production value”, the budget of every active production spread across the length of its shoot, running at about 87% of the October 2022 peak. When we published the Six Market Forces thesis we projected roughly 80% by this timeframe, putting the industry’s recovery “ahead of forecast”, which isn’t something I get to write often lately.

The number that carries the texture is the average production value, which was $15M in July against recent averages nearer $19M, with the count of new productions up 14% on July 2025. More productions, made for less money each. That’s not a failure of anything. It’s the cost compression we forecast in the Six Market Forces work, showing up close to the timeline we expected.

Some of the gap between that number and the feeling is geography, and the redistribution is sharper than the aggregate lets on. New York took 46% of all the new US productions we tracked in July, which is an all-time high in our data, and its share for the year has moved to 26% from 18% in 2025. New Jersey’s smoothed value last year ran 60% above the 2022 peak while the US as a whole sat 35% below it. Measured by production budget absorbed per active sound stage, the UK is running at $0.79M against California’s $0.45M. If you’re on a stage in New York or New Jersey this year, or running post there, where the standalone credits have done exactly what they were designed to do, the recovery feels exactly like a recovery.

But geography explains where the work went. It doesn’t explain why so many of the people doing that work aren’t making money doing it.

Three clocks

Busy and profitable came apart this cycle, and they came apart for three separate reasons at once. The money that funds production changed shape, and the new shape carries less of it into content. The capital vendors acquired to deliver that work became obsolete faster than they could amortise it. And the tools underneath the whole business are being rebuilt while everyone is still using them.

We’ve been through transitions before and mostly survived them, and I think what made them survivable is easy to miss. Film to digital. HD to 4K and HDR. In both cases you knew where the transition ended — and a predictable endpoint is what turns a capital purchase into a business decision rather than a bet. You could model the payback, buy the room, amortise it over seven years, and be reasonably confident the room would still be the right room by the time you’d finished paying for it. That predictability is why good operators kept investing through those transitions, and why most of them came out the other side.

None of these three transitions is revealing a clear destination. And they run on clocks that don’t line up: the business model on a multi-year cycle, the production process on something like eighteen months, the tools underneath feel more like a quarter in innovation cycle time.

One: the money arrived, and went somewhere else

Jack Rogers (Sohonet’s data scientist) has a piece coming for SPI subscribers on the streaming cohort. For the first time, the entire US streaming cohort is profitable. Every major service that reports the figure is in the black; Apple Studios and Amazon Studios don’t break out the number. Peacock was the last to get there, posting a first-ever quarterly profit of $189M against a $101M loss a year earlier. Netflix is running $4.2B of quarterly operating income at a 33.4% margin, which is the win it looks like: the company that committed to the model earliest has the best margin in the business. Disney’s streaming segment more than doubled to $712M, Warner Bros. Discovery posted $512M, while Paramount Skydance’s direct-to-consumer business made $366M.

However, if you read the shareholder letters, not one of those studios names content investment as the growth engine. Every one of them names advertising. Netflix expects its ads revenue to roughly double to $3B. Around 40% of HBO Max subscribers are now on the ad tier, up eleven points in a year. Paramount+ ad revenue grew 30% in the quarter. Where content investment does appear, it appears alongside cost discipline: Paramount reported per-episode production costs down 10% year on year, and Disney named labour and SG&A reductions as an explicit lever.

The working assumption has been that profitability comes first and content investment follows. Those two aren’t sequential. The profit came from cost discipline, so it frees nothing up.

This is where it stops being a story about decline. On the IAB’s numbers, digital video has gone from 29% of US television and video ad spend in 2020 to 58% in 2025, and will pass 60% for the first time this year, with US digital video spend heading past $80B and growing about 20% faster than the total ad market. Inside that, social video has overtaken connected TV for the first time, $28.2B against $26.5B last year, and pulling further ahead this year. Hallen’s framing was the right one: the money didn’t disappear, it changed hands, and new owners bring new margins, new supply chains and new content production chains. And the buyers have changed what they’re buying on: the IAB now has targeting overtaking content quality as the top criterion for television and video investment, up ten points in a year.

So there’s more video being made than ever and more advertising money behind it than ever. - - (Let that sink in for a moment.) - - However, a lot less of it runs through the traditional Hollywood stack. It’s a growth story — it’s just not the growth story Hollywood would have written for itself.

It would be wrong to read this as “the studios” prospering while everyone below them struggles. Paramount starts accruing a $7M-a-day ticking fee on 1 October for a Warner deal it can’t close ahead of an antitrust trial set for March 2027, which is something like $1.2bn by the time a verdict lands, if they don’t negotiate a settlement earlier. Several of the studios have been conducting RIFs as they work through their own cash generation and balance sheet challenges created by the macro changes to the distribution landscape (linear TV vs. digital, etc). The vendor feels this last only because it sits at the end of the cash flow.

Two: the process moves faster than the amortisation

The second transition is quieter, and it strands capital.

The shape will be familiar to anyone who’s bought equipment. A technique unlocks a new service. You invest the capital to provide it — the room, the hardware, the fixed labour — on a five to ten year amortisation. Then the filmmaking process turns again long before the asset is paid off.

Virtual production is a great example. The practice itself is expanding, not shrinking. What’s stranded is the 1.0 configuration: a large permanent LED volume, high fixed cost, paying back only at high utilisation. And the techniques for getting the same result have improved so rapidly that the capability now comes off a much smaller wall, or none at all. Foundry shipped native Gaussian splat support in Nuke late last year: research curiosity to standard toolchain in about eighteen months. Nobody who built a volume was wrong about virtual production. They just bought 1.0.

And that pattern isn’t specific to virtual production volumes. It’s the grading suite, the machine room and the mixing stage, which is the point I was making last month in *Quo Vadis, Post?* about the room never having been the moat.

Three: the tools underneath

The third transition is the broadest and the least discussed, largely because it’s not about filmmaking. It runs from payroll to plate ingestion: scheduling, budgeting, clearance, QC, conform, delivery, media logistics.

When the cost of doing something yourself collapses, the line between making and buying moves. While that boundary has moved plenty of times before, what’s different now is that it moves inside a workflow rather than around a whole product, and buyers are rewiring individual steps with AI agents as they go. This isn’t speculative in the industry next door: the IAB found two in three digital video ad buyers already live, testing or planning agentic AI this year, with another 28% investigating.

It runs both directions, which is worth saying plainly. The same collapse in cost lets suppliers reach further into their customers’ operations than they ever could. This is a boundary moving, not a boundary being breached. But it does mean a competitive set that used to be stable and countable no longer is, because a step you sell can be absorbed, automated or re-scoped inside a quarter.

Why this one is harder

All three are running at once, none of them has a clear destination, and they move at different speeds. You can’t plan capital investment against that.

And the people hit hardest are the ones who behaved the way the last transition taught them to behave: they bought the room, hired the floor, installed the kit, and amortised over seven years. That’s a legible error, made for good reasons, by capable people, and I wouldn’t want anyone reading this to mistake it for bad management (e.g. hindsight is 20/20).

What the index can and can’t see

Four of our Six Market Forces run straight through this: linear-TV cashflow decline and keeping up with Netflix through the money, AI-driven cost compression and hybrid-cloud economics through the process and the tools. The thesis always said the growth would come from new places and from the lower half of the budget pyramid, with XL and L title counts flat out to 2030 while the M and S tiers compound at around 10% a year. The IAB data is independent confirmation of the mechanism: the money followed the ad dollar into formats and creators outside the traditional system. We present those forces as a list of six, and a list implies they queue politely. They don’t, and what a list can’t show you is what happens when several of them arrive together and interact.

Which brings me to the honest limit of our own instrument. The SPI tracks scripted, live-action productions from major studios and perennial producers. It measures work entering production and it measures that well. It was never a proxy for whether the money behind the work is growing, whether the assets you bought still fit it, or whether the step you sell still exists as something anyone buys. And it can’t see social video at all, which is now the largest and fastest-growing slice of the pool that ultimately funds the wider “video production” ecosystem.

Production volume came back. The business model, the production process and the tools underneath all changed while it did, and those are the three things that turn volume into profit.

If you’re looking at a single number and wondering why it doesn’t match your P&L, I think that’s why. The volume line will most likely keep climbing, and there is more video being made and more money behind it than there has ever been. The useful question is which of the three clocks you’re building against.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

Post-production doesn’t behave that way at all (see previous substack). When you automate a task, it simply goes away, and the revenue usually goes with it. In many of my follow-up discussions over the past month the conversation invariably turned to VFX and how it might evolve.

I think there are two lenses we can use to view the data, and they will help anchor the rest of the discussion:

Creative decision context — how much human judgement sits behind a shot or task before anyone can say it is right. Wire removal, almost none. A hero creature in the third act, an enormous amount.

Creative elasticity — how many cycles of fix and review a shot attracts, which scales with how much it matters to the story. The dragon’s first reveal goes round that loop again and again. Wire removal has an elasticity of zero — it has a definable completion state, and there is no version of more gone.

VFX work has been a fixed-bid business for close to thirty years, so those cycles have always sat inside the bid, and it was labour that absorbed them — make one more change to the shot and the team just works through the weekend, no price change. As the cost of a shot begins to shift from labour to tokens, holding a fixed bid across an open-ended number of cycles gets a lot harder, and creative elasticity probably starts hitting stop points other than the delivery date.

The cascade

A visual effects shot is generated imagery, and economically it does not matter much whether the generator is a render farm, a simulation or a physical model. So AI is not a break from what came before, just the latest thing to make a shot much cheaper to produce.

The pattern is that every time the price of a generated shot fell, the capability dropped a budget tier. 1993, tentpole only — Jurassic Park, under 60 fully-CG shots. 2000s, all tentpoles — The Return of the King at around 1,400 effects shots, Avatar at more than 1,800, thousands of shots becoming normal above $150M. 2015, premium episodic — Sense8 ran roughly 1,200 shots in its first season, about a hundred an episode. 2022–24, episodic at feature scale — Stranger Things S4 at 4,853 shots, Shōgun at a little over 4,000 across ten episodes. 2022–23, sub-$25M features — Godzilla Minus One at 610 shots on a reported $15M budget, Everything Everywhere All at Once at roughly 500 shots with a five-to-seven person core team.

Each step took less time than the one before — a decade, then a decade, then about seven years — and each was enabled by a different cost falling: render, then cheap compute, then cloud plus offshore labour. And the mechanism looks the same every time, which is the bit that matters for what comes next — a tier adopts VFX at scale roughly when the price per shot drops under what that tier can carry. Television did not discover an appetite for effects in 2015. The appetite was always there. What changed is the cost of a shot finally came in under a TV budget’s ceiling, and once it did the adoption was not gradual — Sense8’s hundred shots an episode became Shōgun’s four hundred inside a decade. AI will be doing the execution work that cheaper rendering and then cloud did before it, so I would treat this history as evidence rather than a loose analogy.

The paradox of cheaper deliverables

There is a name for this. In the 1860s the economist William Stanley Jevons noticed that more efficient steam engines did not reduce Britain’s coal consumption, they increased it — cheaper coal simply made more uses of coal worth doing. That is the Jevons paradox: make a resource cheaper to use and consumption tends to rise rather than fall. Compute has behaved this way for 70 years. However, in each technology evolution, cheaper rendering did not produce cheaper films, it produced more shots and more elements inside each one. Non-linear editing did not shorten post, it produced more cuts, more versions, more footage.

The Avatar film series is a good case study of that technology march. On The Way of Water they report 18.5 petabytes stored, 18.5x the original Avatar, and 13.6 million threaded render hours for the single hardest shot in the film. By Fire and Ash it was 140 petabytes, 1.248 billion render hours, 200–250 terabytes a day — against a delivered shot count that stayed flat around three thousand. Storage up 7.6x, finished output flat. That looks a great deal like creative elasticity (magnified by image density expansion): the extra compute and storage capacity did not buy more finished shots, it bought more creative detail and review loops.

However, for the paradox to be true, the cheaper and more accessible creative must have a wider audience. It needs the marginal output to be worth something to many somebodies, and stereo 3D is sitting right there in our own history as a good counter example — capability expanded, titles multiplied, then the whole thing retreated because audiences stopped paying for the extra at scale. So it holds where the extra image is worth something to somebody, and fails where it is not — which is the split the two lenses sort for. It applies where the decision context is deep and the elasticity is high. It does not apply to wire removal, where the context for that creative task is nil, the elasticity is zero, and a second pass buys you precisely nothing. That half is likely to be highly impacted by this technology transition.

That is the difference from post. Post sells tasks that complete — a conform is done or it is not — so automating one subtracts it, and the revenue leaves with the work. That is what hollowed out a stretch of British post inside a record £6.8 billion year of UK film and high-end TV spend. VFX sells generated shots, and making those cheaper does not take work away, it usually buys more shots, complexity and creative ambition. So the dividing line is not post on one side and VFX on the other — it runs through the middle of both.

What AI actually changes inside a shot

While AI is likely to disrupt the majority of tasks inside the visual effects process, the same tools produce opposite results depending on where they land. Almost every part of the process gets better, faster and cheaper — breakdown and bidding, previs and planning, asset generation, matchmove, roto, prep, cleanup, and of course the actual image generation step itself. What happens next depends on how much creative context is required for that task or workflow.

Where the decision context is low and the elasticity is zero — wire removal, plate cleanup, paint and prep — the savings are real and probably permanent. That work behaves exactly like a conform. It gets cheaper, then automated, then pulled in-house by whoever owns the pipeline, and it leaves the vendor market altogether. That half of VFX is on much the same timeline as post and heading much the same place.

Where the creative context required for the creative task or shot is deep and the elasticity is high, I think something different is likely to happen, similar to previous cycles. The money saving gets partly eaten and the time saving gets eaten more or less entirely. Cut the cost of a cycle in half and my expectation is that you do not get the same dragon for half the money, you get more cycles on the dragon, because the loop was never bounded by cost in the first place — it was contained by the calendar. If that holds, a faster cycle does not end the loop earlier, it fits more loops inside the same delivery date, so some of the money comes back to the production and very little of the time does. Nobody publishes cycle counts, so this is a read on the direction of travel rather than a forecast. But given what happened at every previous step change in the cost of a generated shot, it is hard to believe all of the gains from AI turn into cost savings. They did not with rendering, they did not with DI, and they did not with non-linear editing.

More shots is not more money

More shots is not the same thing as more money for the vendors doing the work. Part of it is the familiar treadmill — more shots at a lower unit price, running to stay ahead of a price that keeps falling, which is why you can be busier than you have ever been and still be eating cash as a vendor.

The bigger part is the fixed-bid point from the top of this. Those cycles have always been free to the person asking for them — not because anyone was generous, but because labour is fungible and a metered bill is not. You cannot bury tokens inside somebody’s long day the way you could bury a weekend, so as compute becomes more of the mix, every cycle carries a cost from the moment it is incurred. Which is why I do not think AI solves the fixed-bid problem so much as removes the thing that made fixed bid possible. For a counter point, there is of course the opportunity at some point of scale to leverage internal model hosting capacity (yes, internally hosted capex).

The volume expansion is also conditional rather than assured. The cascade only continues if the tiers underneath actually want the shots once they can afford them. TV wanted them and proved it the moment the price came under the ceiling, and smaller features have always wanted effects they could not afford, so I would expect both to behave the same way again. But appetite is a property of format, not budget — unscripted and vertical drama are faces and dialogue by design, and no price cut creates a demand that was never there. That is the real brake on this argument and it deserves to be taken seriously.

The floor is the decision

Every cost in a shot has fallen — hardware, render, licences, labour via offshoring, now execution — with one exception, which is that somebody still has to say the shot is right. So the minimum viable price of a shot converges on the cost of a single supervision cycle. That is a human-attention number, not a technology number, and it is precisely the price of creative decision context.

Last month I argued that what endures in post is the decision with real context behind it. This is the same claim with an economic floor under it — context is not only what survives, it is what the money gets spent on, because the shots with the deepest context are exactly the shots that attract the most cycles. This trend would predict a tier below the floor, with plenty of shots and very little judgement per shot, which is already a fair description of most AI-generated video today. We just call it slop.

There are two potential outcomes worth discussing. One is that pricing follows the cost structure the way software did — iteration becomes a metered line item, directors get a version budget, and we finally put a price on the thing we have spent thirty years giving away. The other is that nothing changes in the industry’s operating model, vendors absorb the cash cost by building and hosting open weight models internally, while the industry’s vendor mix evolves into a two-tiered market, with a large number of smaller, creative firms rising to meet the higher volumes and a second tier who have the fixed infrastructure to handle tentpole features but are highly exposed to market volatility.

Three of our Six Market Forces run straight through this — hybrid-cloud economics, AI-driven cost compression and aggregation. The shot count and complexity is most likely going up, because it has gone up every time the price came down. What might change is who pays for the extra cycles, and for thirty years that was never the person asking for them.

If you have a different read, let’s discuss.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

The Screen Production Index tracks productions by region and by total budget, and that budget spans the whole envelope: the camera work on the floor, the post, and the visual effects, through to delivery. Those stages used to rise and fall together by region. This month they didn’t, and the gap is the point.

What a post house actually provides

Strip away the mystique and a post house does five things: it brings the craft talent (the colourist, the mixer, the editor); it provides the purpose-built facility (grading suites, screening rooms, machine rooms); it runs the conform that moves work cleanly between tools and workflows; it provides asset management, holding and tracking and moving the material; and it does the wrangling, the project management (or throat choking) that brings every vendor and workflow to a finish, on schedule. For thirty years those five sold as one bundle. What’s happening now is that the bundle is being priced apart, function by function, and the market is paying very differently for each.

The old logic

Rewind to the pre-strike world of 2021–2023. Back then post production broadly followed production. Put the money and the cameras in a region and a predictable share of the finishing followed, with a standing tilt toward California, because that’s where the studio post executives sat and habit kept the work. Visual effects ran on a different but equally legible logic: facilities chased the long-dated tax credits, invested in physical capacity where those credits lived, and ran their own follow-the-sun pipelines across London, Vancouver, Mumbai and Sydney. You could draw the map. Production value in a territory was a decent proxy for the post and VFX value that would land there.

The recovery came back, but not the way it left

Total production value is recovering (to roughly 86% of the 2022 peak, running ahead of the ~80% our Six Market Forces thesis projected). What’s changed is how. I wrote last month about one face of it: the tax credit has been unbundled from the location. Standalone post and VFX credits now stack talent plus incentive in specific places, and the two that stacked them highest are New York and London. New York’s smoothed production value is up 178% year-on-year in our latest data; the UK posted a record £6.8 billion in 2025 film and high-end TV spend, up 22%, on fewer titles. On the old logic, the post ecosystems in those two cities should be booming.

The paradox

They are not. Our July SPI puts a fine point on how hard the UK is working: it is absorbing $0.79M of production budget per active sound stage, against $0.38M in California and $0.44M in New York, roughly twice as busy. And yet the last eighteen months have hollowed out a stretch of British post. Technicolor took MPC, The Mill and Mikros down in early 2025. Glassworks wound up. Marshall Street Editors closed after 27 years. Absolute and Blind Pig shut in May; No.8 entered insolvency in June; the group trading as Halo, Evolutions and Relay went into administration over the winter. A single closure is a company story. This many respected houses failing during record production spend and the world’s busiest stages is a market story, and it means the value is tearing along some seam other than volume. So where?

The seam: automate-first vs. the residual

The useful move is to stop treating a production as one thing and see it as a stack of tasks, then sort those tasks on two axes: the stakes (what does it cost if this goes wrong?) and the verifiability (can a machine confirm it was done right?). The two pull in opposite directions in a pretty obvious manner.

One pile of workflow tasks is automate-first, and it maps straight onto high activity lines at a post house: conform, transcode, packaging, and the holding and moving of media. High-volume, machine-verifiable, low-drama, and its price is being commoditized by software on one side and by studios pulling the work in-house on the other. This is the layer under the most pressure and the hardest to keep charging for. It is also, not coincidentally, most of what a traditional post house quietly bundled into the price packaged with talent. The London vendors weren’t short of work. Too much of what they sold was automate-first work, and that’s exactly the kind of work the market is driving towards automation.

Los Angeles, for contrast, is the simple case. There the pain is plainly a volume problem: FilmLA shoot days fell 16% in 2025, television is off nearly 58% over three years, and stage occupancy that held in the mid-90s has slid to 63%. The work physically left on incentive math, which is why California rushed a $750M credit expansion to claw it back. LA is losing the work. London is doing the work and its vendors are failing anyway, because a large chunk of the tasks being automated away are currently included in the post revenue model.

Where the value is leaking

Watch the buyers and you can see the strands being pulled. Studios worried about the fragility of their post supply chain are building finishing capacity in-house. Sony is winding down Pixomondo and folding the capability into its own Imageworks, and several are routing productions through their own cloud tool-chains, absorbing work that used to sit with an external post house. The studio that frets about post-house fragility and responds by internalizing the workflow is, without meaning to, pulling away the very work that would keep that supply chain alive. Meanwhile the technology has flipped the economics from capex to opex: modern, largely AI-assisted tool-chains let a handful of people take on jobs that used to need a floor of them, and because the work is cloud-native, teams in lower-cost places can service the classic workflows for anything below the top budget tier. The room stops being the moat. The value doesn’t vanish; it migrates to the parts of post that a studio can’t easily internalize.

What doesn’t automate, and why

Here’s the other pile of post workflows. What resists automation is the work where the stakes are high and “right” can’t be checked by a machine: the creative decisions. And the value there isn’t the task; automation can execute the task. The value is whose context sits behind the decision. A director keeps control because the call is theirs and the context — what the film is meant to feel like, what note the audience should leave on — is human and can’t be handed off. A colourist or a sound lead endures because the client trusts their judgement, not because the tool is rare; the moat is the accumulated read on what this director wants. Craft with thin context around it — format conversion, packaging, asset shepherding — is the exposed part. Craft with rich context behind it is the durable part. The wrangling, the fifth function of bringing every vendor to a finish on time, endures for the same reason, but its value scales with complexity: on a sprawling production with dozens of vendors it is worth a great deal; on a small one there is far less to coordinate, and even that thins.

The bigger picture: three of the Six Market Forces

Our Six Market Forces framework maps how film and TV production reshapes between now and 2030: the structural shifts remaking the business. This one change shows three of those forces arriving at once. Hybrid-cloud dissolves the advantage of owning the room. AI-driven cost compression resets the labour math a facility was built around. And aggregation (studios and platforms pulling the workflow in-house) moves the coordinating role toward the buyer. One discipline worth keeping honest as this plays out: what automates is a different question from how big the market is. Automation reshuffles who does the work; the size of the prize depends on value-per-title, and with the premium pool capped and the volume pool exploding, those two curves are not symmetric. More content does not mean more revenue.

Why it doesn’t reverse

The uncomfortable part for the existing post base is that the trend compounds. The pool of genuinely large-budget work is effectively capped (there is no world with materially more tentpoles than we have now, and our own tier data already shows the extra-large slate running below its 2022 peak), while the studios that command that work increasingly fill their own capacity first, stages and post alike. So the incumbent is squeezed from both ends: the high-budget work that sustains a big fixed-cost facility is finite and increasingly captured internally, while the growing mid- and lower-budget work is exactly the segment the new opex-light, distributed model serves best. None of which means post is dying, and not everyone accepts the sharp version of this: above a certain budget line the finishing on a marquee title still goes to a reputable tier-one post house, and for years it will.

The half that endures

For thirty years, a post house won work by answering one question: do you have the talent, the creative workflows and the purpose-built facility? That second part of the answer (the machine automatable workflows, the grading suites, the screening rooms, the machine rooms) is losing its value. The automate-first work no longer pays well enough, and the facility itself was never the real advantage. What endures is the other half: the talent, and the trusted context they bring to the decisions a machine can’t make. The SPI measures where productions happen; it can’t see a post house’s balance sheet. But the stage-usage number is the first place the decoupling becomes legible, and it’s telling us the question has moved from how much is being made to who holds the context behind the work, and who captures the value for holding it. That’s not the death of post — it’s the evolution of what post is for.

I have been thinking about how to write this blog since last summer, was inspired by a Televisual dinner with post execs in London recently, and a big thank you to Doug Shapiro for helping me find the right frame for the changing value trends over a coffee.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

Last month I wrote about one piece of this — AB 2319, California’s attempt to win its post-production work back with a standalone rebate. That was the view from the statehouse: a jurisdiction trying to claw a slice of the value chain home. This month’s data shows the other side of the same coin, and it’s the side that actually matters to the people reading this report. Because what looks like a fight between jurisdictions is, from where a producer sits, something more useful: more flexibility in how you build a production.

The data underneath the headline

Start with what’s solid. New York’s share of US smoothed production budget has climbed from 12% in 2022 to 17–18% across 2023–25 and through 22% in early 2026. That gain is spread across the traditional broadcasters and the streamers — not one tentpole flattering a month — with 47 productions in the forward pipeline. The SPI attributes the move plainly: it reflects “the highly competitive tax credits now in place.”

A caution worth keeping honest: the month-to-month regional split is noisy. The UK took 29% of new productions in April and just 16% in May; the US swung back to 46%. One month doesn’t make a trend, and the SPI says as much — two more cycles will tell us whether May’s US dominance is structural or wobble. New York’s gain is more trustworthy precisely because it’s measured on an eight-month trailing basis, not a single print. The skyline didn’t get more photogenic. The incentive got more generous and specifically, more generous about post production. New York runs a 30% refundable post-production credit, funded at $45M a year through 2036, sitting inside a state film program whose cap was just lifted to $800M, the highest in its history.

The credit grew up

Here’s the shift that I think defines the next two years. For most of the last decade, a film tax credit was a single instrument tied to a single thing: where you shot. You picked a location, you qualified your spend there, you collected. Post, VFX, and finishing came along for the ride or didn’t, mostly following talent and habit rather than money.

That’s no longer how the map works. The credit has been unbundled from the location, and it’s happening in parallel across jurisdictions. California’s AB 2319 (still working its way through the Assembly, so watch the floor vote before banking it) would create a standalone 35-to-50% post-production rebate that explicitly does not require principal photography in the state: shoot anywhere on earth, finish in California, still qualify. Georgia revived their 20% post-production credit in January of this year, with an additional 10% if you also shoot there. The UK lifted its visual-effects credit to an effective 39% rate, removed the cap on qualifying VFX spend, and made generative-AI VFX costs eligible. Ireland pushed its VFX credit to 40%. Five jurisdictions, each now bidding for a phase of the work — finishing, VFX, audio — decoupled from where the cameras rolled.

What that gives a producer

Put yourself in the chair of someone financing a large feature. A decade ago the credit math pushed you toward one answer: find the jurisdiction with the best overall package and concentrate the production there. Today the same math hands you a menu. You can shoot where the location and the production credit work hardest, route the visual effects to wherever the VFX uplift is richest, and land finishing where a standalone post credit tips the budget — and each of those can be a different city, or a different country.

That is the real meaning of New York’s 178%. It isn’t that producers suddenly prefer New York. It’s that the credit map now rewards splitting a production into pieces and placing each piece where it pays best — and New York built an attractive piece. The large feature is increasingly assembled, by design, with a global footprint: production in one place, VFX in another, post and finishing somewhere else again, each chosen on its own merits.

The part that doesn’t show up in the credit math

A distributed production is a more flexible production, but it is also a harder one to run. Every seam between those locations — every handoff of dailies, every VFX pull, every conform and review cycle that now crosses a border — is a place where schedule, version control, security, and creative intent can leak. The flexibility is real; so is the coordination tax that comes with it. This is the same fragility I flagged last month in a different guise: efficiency you’ve unlocked on paper only shows up in practice if the operational layer underneath it can actually carry the work across the map.

So the producers who turn this flexibility into an advantage will be the ones who treat the global footprint as a capability they’ve built on purpose — everything required to keep a production coherent when its phases live in four time zones — rather than a logistical accident they’re absorbing job by job. The credit decides where the work goes. Whether the work survives the journey is a separate question, and it’s the one that’s quietly getting more important.

The caveats are real

Incentive regimes run in feast-and-famine cycles. The same lever that opens a lane for New York or California can be outbid or unwound the moment another jurisdiction moves or an administration changes. AB 2319 is not law yet. And the UK’s softer month is a reminder that today’s structural gain is tomorrow’s competitive target — which is exactly why the BFI is lobbying to keep UK screen incentives walled off from any Trump-era trade deal. The map is not settling down. It’s getting more granular.

What’s new this month

The SPI is a lagging indicator by design. It will show us in 2027 and 2028 — through the smoothed budget and the post-production mix that trails productions to delivery — whether this fragmentation hardened into the normal way large features get built. The leading indicator is already here, in a 178% that isn’t really about New York at all. The credit used to ask producers one question: where will you shoot? It now asks a better one: where will you do each part of the work? That’s more freedom for the people building productions — and more to coordinate for everyone who serves them.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

The cost compression itself is real, and it’s accelerating. In our Six Market Forces report, we named Generative-AI Cost Compression as Force 1, the first force reshaping production through 2030, and the receipts are now arriving. At the Cannes market in late May, Higgsfield showcased Hell Grind, a 95-minute AI-generated feature reportedly made for roughly $500,000, with about $400,000 of that going to compute. Days later, Tribeca accepted Ash Koosha’s Dreams of Violets, widely described as the first fully AI-generated live-action feature accepted into a major festival’s official lineup. The compute-heavy economics and the institutional acceptance arrived in the same fortnight. Runway’s CEO Cristóbal Valenzuela framed the prize plainly: AI could let Hollywood “make 50 films instead of one $100M blockbuster.” The money is moving from people, sets, and locations to metered compute, and the chain of suppliers in between is getting shorter.

The upside is real — and it isn’t only price

A media supply chain leaks value at every handoff. Creative intent, version history, and the reasons behind a decision drain out each time an asset moves from originator to vendor to sub-vendor. Done well, AI carries that context across the handoff instead of losing it, and compresses the iteration loop so teams try more ideas and reach a better answer faster. Better context and quicker outcomes are genuine gains, and I don’t want the risk argument to drown them out.

But they are conditional. They only show up where the work underneath was already structured so an LLM or agent can act on it: where demand, task, and state exist in a system, not just in a person. Most of it wasn’t.

The chain was never as automated as the org chart implied

That’s the argument David Sugg makes in “When the Human Layer Is the Control Layer,” published last month on Andy Beach’s Engines of Change, and it’s the clearest thing I’ve read on this all year. Sugg spent 26 years building and operating media supply chains at major studios, so he’s describing the plumbing from the inside. His point: in a typical media supply chain, the work that controls, integrates, and orchestrates the workflow doesn’t live in the systems. It lives in the heads, runbooks, and habits of the people doing it. He calls the pattern Human Automation — what you get when an organisation decides, over and over, that putting a person between two systems is cheaper than building the integration. The dashboard reports green because the people kept it green.

He pairs that with an uncomfortable number. MIT’s August 2025 “GenAI Divide” study found that 95% of enterprise GenAI pilots failed to deliver measurable ROI — the reported problem not primarily model weakness but poor workflow integration, which maps directly onto Sugg’s point about systems that never captured the real work. His rule, earned over those 26 years, is the line I keep coming back to: you should never tool a bad process. Agents don’t make that rule less true. They make it more expensive to ignore.

So “more fragile” is two problems, not one

The first is the obvious one. As cost migrates to compute, the supply chain concentrates onto a handful of GPU clouds and foundation-model vendors. When CoreWeave announced the first bring-up of NVIDIA’s Vera Rubin NVL72 on June 1, the claimed economics were stark: up to a tenth of the cost per million tokens compared with Blackwell. That deflation curve is real, and also a reminder that your render pipeline now leans on a few suppliers whose hardware carries lead times measured in quarters, not weeks. The old chain had redundancy through fragmentation. The new one has a small number of very large single points of failure.

The second is the one Sugg surfaces, and it’s deeper. The links that look most industrialised — mastering, localisation, QC, delivery — are often the ones held together by undocumented Human Automation. Point a cost-compressing agent at them and it inherits the runbook with none of the reasoning behind it, then executes that imperfect understanding at speed. The senior tier is worse: the routing knowledge in one PM’s head, the Tuesday phone call about which vendor is over capacity, the talent preferences in a notebook — none of it is in any system for an agent to inherit. And both buffers that used to absorb this are thinning at once. Internal headcount is being cut, and the outsourcing escape hatch just proved its own fragility when Technicolor, 110-plus years old, was liquidated globally in early 2025, leaving in-flight work scrambling for a home.

Both of those fragilities are amplified by a property of the tools themselves: LLMs are non-deterministic. The automation our supply chains were built on — transcode, checksums, conformance, delivery — is repeatable by design. Run it again, get the same result; that repeatability is the basis of QC and of the trust between suppliers. LLM-driven steps break that assumption. The same input won’t reliably produce the same output twice, so a step that passed yesterday can fail today for no visible reason, and “green” stops being a guarantee of anything. That’s not a reason to avoid these tools. It’s a new property the operational posture has to manage, with validation, guardrails, and human checkpoints designed in rather than bolted on after a bad delivery.

Whose agents run the supply chain?

This is the shift I think matters most, and we flagged it just weeks ago. After the Devoncroft Summit ahead of NAB this April, I wrote that Josh Stinehour had put the iteration-cost question on the main stage, and that the room kept circling a bigger idea: operators and content originators want to push their own AI layer into their vendors’ tools, making themselves the orchestration layer and treating each vendor tool as a service they call. In that world, the user of a vendor’s tool isn’t a person at a seat. It’s an agent, running thousands of actions a day, often unattended, often alongside other agents. The originator at the head of the chain stops buying tools and starts orchestrating across suppliers with agents of its own.

That inverts the supply chain, and it runs straight into Sugg’s wall. An originator’s agent can only orchestrate a vendor whose demand, task, and state are exposed in a form an LLM-driven agent can act on. So the question Sugg says every operator should ask itself — where does your in-flight work actually live, and if the honest answer is a spreadsheet, an email thread, or a senior PM’s memory, that’s your foundation gap — becomes the question an originator’s agents will force onto every supplier in the chain. A vendor that’s a human-automated black box can’t be a node in someone else’s agentic workflow. It gets routed around.

The human layer can also be operationalised on purpose, before it breaks. While Higgsfield’s louder peers chase the cheapest possible feature, the Terrarium project from Jason Zada and producer Steven Schneider, financed by Artlist, took the opposite tack — building its AI-hybrid production as a fully DGA- and SAG-AFTRA-compliant workflow. That’s the human layer written into the contract rather than left in someone’s head: a deliberate decision about who carries the context, who decides, and who’s accountable when the agent is wrong.

Where this lands for the DPP — and for the SPI

Here’s where that leaves the rest of u: the dividend is real and large, but you can only bank it where demand, task, and state already live in connected systems — and across most of the chain, they don’t yet. That’s not a failing to atone for; it’s simply where the industry’s capital went for twenty years. The forward-looking point is the one I’ll be making in New York: the unglamorous work of making supply-chain state legible and interoperable across suppliers is now the thing that decides whether the AI payoff actually lands. That’s the work the DPP has been organising around, which is why it’s the right room for this conversation.

Our Screen Production Index is a lagging indicator by design; it will show the volume effects of all this in the budget tiers long after the headlines. The leading indicator is already here, and it isn’t the price of a shot. It’s whether your supply chain — and your vendors’ — is structured well enough that an LLM-driven agent could actually run it.

AI doesn’t automate your supply chain. It automates the parts you already operationalised. Everything else is still a person you’ve worked with for twenty years. The efficiency is real. So is the fragility underneath it. The question for everyone in the chain is which one you’re building for.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

That’s worth sitting with for a minute. What AB 2319 proposes is different from what the rest of the California tax-credit conversation has been doing for the last twelve months.

Two California credits, two different problems

AB 1138, signed last year, expanded the existing film and TV production tax credit. It more than doubled the annual cap from $330M to $750M, raised the base credit to 35%, and added 40–45% uplifts for filming outside the LA zone and for VFX work. Since passage, the state has seen a roughly 53% year-over-year increase in productions approved for the incentive. Confirmed in-state productions now include sequels to The Simpsons Movie and Everything Everywhere All at Once. That’s the production credit. It targets the line items that show up in our SPI’s New Productions count: principal photography, crew, stages, locations.

AB 2319 targets something different: post-only work. Editorial, colour, sound, finishing, VFX. The line items that don’t trigger the production credit because they aren’t shot in California, but where the work itself can absolutely be done in California, and for the last fifteen years, has increasingly not been.

There’s a workflow reason this matters more than the credit math suggests. A meaningful share of the senior post talent — editors, supervising sound editors, finishing colourists, VFX supervisors —along with the post executives from the studios are based in Los Angeles, regardless of where a project shoots. For years, productions shooting in Atlanta, the UK, Vancouver, or Eastern Europe would still bring picture editorial, sound, colour and finishing back to LA, because that’s where the key creative collaborators worked and lived.

The lack of a standalone California post credit has steadily eroded that pattern: when the production credit is the only lever, the rational choice is to do post wherever the production happens, and ship the LA-based creative supervisors to the shoot location instead. AB 2319 puts the option back on the table, productions can shoot wherever the production credit math leads, and bring post back to where the talent already is, without giving up incentive value.

Tim Belcher flagged the bill to me back in April, alongside the broader work the California Post Alliance is doing to keep post-production employment in-state. The bill was still being introduced then. Six weeks later, the appropriations confirmation arrived. Not yet law. It still needs a floor vote and the governor’s signature, but the appropriations hurdle is where most California spending bills die, and this one didn’t.

The April data, read against the legislation

Here’s where the SPI numbers get interesting.

April was a weak month for the US in our data. The US share of new productions tracked dropped from 46% in March to 30% in April. Within the US, California’s share collapsed from 46% to 21%. The UK doubled its global share from 16% to 29%. Georgia bucked the trend and climbed back to 10% of US smoothed spend, its highest reading in over a year.

In isolation, that 30% US month reads as cyclical lumpiness. The SPI tracks principal photography starts, and a single month of slate concentration in the UK doesn’t reverse a recovery. The Q2 forecast is still +35% on Q2 2025 across all tracked geographies. The medium-term direction is intact.

But the policy stack underneath the cyclical noise is shifting in a way that matters more than April’s single data point. AB 1138 is already showing up as a 53% YoY increase in approvals. AB 2319, if and when it clears, gives the standalone post pipeline a credit lever it has never had, with a six-to-nine-month lag behind whatever the production credit clawbacks deliver. The hubs that built post capacity in London, Mumbai, Bucharest, and Vancouver over the last fifteen years did so by serving real demand under a stable incentive structure; AB 2319 doesn’t change their fundamentals, but it does introduce a competing lever where one didn’t exist before.

That two-step policy mechanism — production credit first, post credit behind it — is the most consequential change in any single jurisdiction’s industry policy since California’s original tax credit launched in 2009.

The trend AB 2319 is built to reverse

LA-based post houses have been closing at an alarming rate. The global Technicolor Group’s collapse in February 2025 took down its LA operations along with the rest. A string of small post and VFX vendors have closed and a number of larger service providers slashed the staff in their LA facilities. Light Iron was shut down by parent Panavision earlier this year and only re-emerged in mid-May after a management-led buyout took the brand and the Hollywood De Longpre facility out of bankruptcy and relaunched it as an employee-owned outfit, a redemption story, but only after the original entity went to the wall.

None of those closures was driven by a lack of production demand. They were driven by where the post work was going, and the work was flowing, rationally, to wherever the credit math and the production location together made the most sense.

This is the contraction AB 2319 is structurally built to slow or reverse. The bill won’t bring back what’s already closed. But if it lands, it changes the unit economics for what’s still standing, and it changes the equation for anyone considering whether to keep a Los Angeles post facility open through the next slate cycle.

Force 2 of our thesis, with a post-shaped twist

This is Force 2 of the Six Market Forces forecast we published in January continuing to disrupt production decisions: geography reshaping. In my last blog we discussed overall project volume including more XL and L productions than we had anticipated, but it would be the production credits taht determine where they would be shot.

What the production-credit version of Force 2 misses is the part of the map that’s downstream of the shoot. Post-production work has its own credit map, its own labour pool, its own infrastructure costs. Over the last fifteen years, that map redistributed: Mumbai built world-class VFX capacity, London anchored picture finishing for major tentpoles, Vancouver, Toronto, and Atlanta picked up editorial and sound work that followed the productions there. None of those hubs are going anywhere: they serve real demand. What changes if AB 2319 lands is that the option to do post somewhere different from where the production shot becomes a viable financial choice again, not just a creative-talent one.

If the policy lands, the SPI counts in 2027 won’t tell you the whole story. The smoothed budget will — and the post-production budget mix that follows productions through delivery in 2027 and 2028 will tell you how the post pipeline actually redistributed. If the bill passes, it will provide more flexibility for productions to be cost effective and below the line talent friendly.

The caveats are real

Incentive regimes create feast-and-famine cycles. The same legislative mechanism that opens a competitive lane for California can close one elsewhere, and can be unwound the moment another jurisdiction outbids or a new administration cuts the program. The UK’s Audio-Visual Expenditure Credit holds at 34% baseline through 2031 with a 39% uplift for VFX expenditure. Canada and Australia remain uncapped and effective in the 40%+ range. None of those programs is going to absorb a California post-credit launch quietly, and the next round of competitive responses is the part of the cycle worth watching and analyzing.

There are also macro overhangs that haven’t cleared. The SAG-AFTRA TV/Theatrical contract approved by the National Board on May 11 is in member ratification through June 4. If members ratify, the May SPI’s slight Q2 forecast trim — from +38% to +35% YoY — likely reverses as paused projects release. If they don’t, the +35% becomes the ceiling. The WBD-Paramount Skydance combination, with EU phase 1 review now cleared, is targeting a July 15 close. That’s optimistic — California’s Attorney General is actively studying the deal — but if it lands, there is risk that the combined entity will rationalise overlapping IP and the XL and L production budget count on our forward calendar gets thinner in 2027 or 2028.

What to watch for between now and Q4

If you’re part of the core audience of this blog — a studio CTO, a post-house owner, a finance team at a vendor that serves either — the questions worth carrying out of the May SPI are important.

First, does AB 2319 clear the floor vote and get signed? If it does, what’s the budget cap and the eligibility window and when does it go into effect? Second, does the May UK share spike at 29% sustain through Q3 or revert to the 22% trailing average? That tells you whether the next round is active competition or structural inertia. Third — for the people who run the production calendar — what does the post-production budget mix on the back half of the 2026 slate look like compared to 2024, and how does it shift if a standalone post credit becomes part of the math?

The production credit alone has never been enough to bring post-production work back. The post-only lever is the structural piece that’s been missing — and the one every other jurisdiction will now be measuring against.

The trend is still up. The map is still wider. What’s new this month is that the post pipeline might soon have a credit lever of its own for the first time in California.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

And the places it’s landing isn’t where it used to be.

What our forecast actually said

Earlier this year, we published our own production industry forecast through 2030 — Sohonet’s Six Market Forces thesis. The headline was simple: total volume grows, but the growth lives below the Large tier. Streaming spend keeps rising 4–6% a year. Creator-economy content keeps lifting the small end. Medium ($10–35M) and Small ($1–10M) titles expand by roughly 1.6x over five years. And for Large and Extra Large? The prognosis was direct: “XL and L budgets likely stay flat in count, but overall pie expands at headline CAGR.”

Flat through 2030.

Then the April reading from our Screen Production Index (SPI): 7% more Extra Large starts than the same period in 2025, running at 78% of the Q1 2022 peak — the last pre-strike comparable. One month doesn’t reverse a five-year forecast. But 51 XL plus 22 L starts already booked from now through year-end isn’t a forecasting artifact. It’s a concrete set of decisions that major studios have already made. The checks are cut. The crews are hired. The stages are leased.

If that’s the what, the where is where the story actually lives.

The map has widened

The slate isn’t converging on one production hub. It’s spreading. California’s AB 1138 raised the annual film tax credit cap from $330M to $750M — a more-than-doubling of capacity that began accepting applications this year. Georgia is uncapped and reinstated its post-production credit. The UK’s Audio-Visual Expenditure Credit regime is locked in at 34% baseline through 2031, with an additional credit for VFX expenditure now operational. Canada and Australia remain uncapped at effective rates north of 40%. Several European jurisdictions have added or expanded film and high-end TV programs in the last eighteen months.

Each Large or Extra Large greenlight in 2026 is also a jurisdiction decision. Where the credits are richest is where the work goes — and the credit map is richer in more places than it has been in fifteen years.

This is Force 2 of our forecast arriving on schedule: geography reshaping. The aggregate volume looks healthy. The geographic distribution is what changed.

What’s behind the willingness to commit

Studios are committing because they have capital to deploy and a stable policy backdrop to deploy it against. Netflix is holding at roughly $17B for 2026, in line with 2025. Apple TV+ is up approximately 6% year-over-year past $2.5B. Amazon MGM has scaled theatrical from 8 releases in 2025 to 14 planned for 2026 — a near-doubling of slate. Disney is guiding to $24B in total content spend across Disney+, Hulu, and ESPN+, up roughly $1B from 2025 and reversing the belt-tightening of the 2024–25 cycle. Paramount/Skydance, freshly merged, is pushing hard — March’s SPI data shows their budget commitment running at 211% of March 2025. Not a typo.

If you’re a studio greenlighting a Large or Extra Large title this quarter, the policy backdrop is more predictable than it has been since 2019, and the demand signal — from box office and from streaming — is the strongest it’s been since the strikes. The slate is the proof.

It may not feel like recovery where you live

Geographic divergence is the trade-off in this trend. The aggregate is healthier. The local picture varies by postal code.

Roughly 17,000 Hollywood jobs disappeared in the last year. California’s expanded tax credit is reclaiming share that had moved to Georgia, the UK, Canada, and Australia — but reclaiming it from a smaller base. If you’re a crew member, vendor, or facility operator in a jurisdiction that’s losing work to a richer credit, the headline numbers don’t help you. Your particular neck of the woods may be in the part of the map that’s getting smaller.

The full-year box office picture is recovering but not recovered. Gower Street and Bruce Nash are forecasting $9.75–$9.8B for 2026, still approximately 22% below 2019’s $11.4B peak. Cord-cutting is running at roughly 5% annually. Linear TV is in secular decline outside live sports, now at 12.4% of global media spend versus 41.3% in 2013. The production calendar is real, and it coexists with an ecosystem that hasn’t finished its contraction.

There is a structural asterisk on the slate as well. WBD shareholders approved the $110B Paramount/Skydance combination on April 23, with a target close of September 2026 — pending US and European regulatory review and an active California antitrust look. David Ellison has publicly committed to a 30-feature theatrical slate per year post-close, doubling down on that pledge at CinemaCon in April. The historical precedent gives some pause: when Disney closed its $71B acquisition of Fox in 2019, the combined entity ultimately released roughly half as many Fox-label titles as the standalone studio had been producing — despite similar pre-close commitments to maintain the slate. Post-merger rationalization is the historical norm: overlapping tentpoles get cut, mid-budget projects get reviewed, slate count shrinks before it rebuilds. Some subset of the Large and Extra Large commitments on the calendar today is at risk of being reprioritized between now and 2027’s actuals. My personal view is that they will maintain volume as long as performance meets expectations -- and I do not think that will be assessed or debated until 2028 or later.

And the same incentive mechanism that’s pulling productions into California, Georgia, the UK, or Vancouver can pull them away the moment another jurisdiction outbids or a new administration cuts the program. Incentive regimes create feast-and-famine cycles. The current map is wider than it used to be in part because the incentive backdrop is stable. That stability is not permanent.

What the conversation looks like in December

Here’s the question worth holding for the rest of the year: if the twenty-two Extra Large and fifty-one Large productions currently on the calendar actually make it through principal photography, through post, and into delivery windows — does the production map continue to widen, or does another jurisdictional shift redraw it again?

Our forecast says the long-run shape of the industry is more content, produced from more places, at lower cost, with distributed teams. The April SPI data may or may not turn out to be the signal that L/XL volume is recovering earlier than projected. Either way, the place to watch is the monthly production data — not in a single quarter, but in the run rate from here forward, and in the geographic spread of where each new start lands. That’s what the Screen Production Index is for.

The trend is up. The map is wider. That’s a better place to be standing than we were at this time last year — and where you’re standing on it matters more than it did before.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

Walking the show floor last week, that question kept surfacing—not in the keynotes about codecs and transmission, but in booth conversations about production workflows. It came up in storage vendor demos, in AI tooling pitches, in side conversations between sessions—different people, different corners of the floor, every day of the show. It isn’t really a pricing question. It’s a business model question hiding behind a pricing question.

What are you actually paying for?

I spent time at a LucidLink demo and asked about their Business plan—$32 per member per month, 400 GB of shared storage per member. I did the arithmetic walking to the next booth: that’s roughly $80 per terabyte per month for a mounted filesystem that feels local. Suite Studios prices in a similar range, and last week announced its file-streaming technology is now embedded inside Frame.io Drive—Adobe’s new desktop app that mounts cloud projects as local. Same wrapper pattern, different distribution channel, and one that puts Suite Studios directly inside the toolchain millions of editors already use.

Look under the hood. Amazon S3 Standard runs $22–23 per terabyte per month at entry-scale tiers. Wasabi charges $6.99 per terabyte with no egress fees. The product these vendors sell is real—conflict resolution, audit trails, tool integrations—and the engineering is good. For a freelancer or a small team, $80 a terabyte is a fair trade.

But the economics shift when you’re a tentpole production generating 500 TB across 50 collaborators, or a scale organization with 500 artists across multiple facilities. At that scale, the spread between the wrapper and raw S3 is roughly $345,000 a year. Against Wasabi-class alternatives, it’s closer to $440,000. Treat those as directional—pricing tiers, contract terms, and storage classes vary—but the shape of the markup doesn’t change, and AI-driven iteration is about to multiply the storage footprint on every active show (along with every other metered infrastructure cost that scales with iteration: GPU and CPU cycles for rendering and inference, egress charges moving data out of public hyperscalers, API calls to AI services).

Where the meter is already running

Once I started looking for consumption pricing on the show floor, I found it everywhere.

AWS Elemental Inference, launched in February and demoed broadly last week, is pay-per-feature and pay-per-minute. Google’s Vertex AI prices generated video at $0.50–$0.75 per second; Gemini inference runs from $0.10 to $4.00 per million tokens. Anthropic and OpenAI were never anything but per-token; their agentic workflows multiply API calls by definition. Even Anthropic just shifted its enterprise pricing two weeks ago—the seat fee no longer bundles token usage, which now bills at API rates on top—a real-time acknowledgment that flat-rate pricing doesn't survive high-volume usage. Adobe Firefly Video, which integrated Kling 3.0 ahead of the show, costs roughly $0.43 per five-second 1080p clip at the Pro tier—the unlimited-generation promotion ends May 20. Cloud GPU rates at the hyperscalers run $3–7 per hour for H100s, around $9.36 for B200s. Custom AI model training inside the new generation of compositing tools is bundled into the seat, but the GPU time to actually run the training—typically $500 to $700 per run on a meaningful dataset—sits on top.

When a per-seat SaaS tool sits on three layers of metered infrastructure, the seat price is doing a lot of accounting work. Light users subsidize heavy users. The vendor’s margin moves with each customer’s usage profile. None of this is a problem at low utilization. All of it becomes a problem when iteration volume rises—which is the explicit promise of every AI feature on the show floor.

The seat that isn’t a person

The structural point came at the Devoncroft Summit, not on the show floor, and it’s the part I haven’t stopped thinking about.

Jon Roberts from ITN stood up and said flatly: “SaaS is dead.” Mike Kralec from Sinclair argued that the agentic AI would come from the broadcast networks themselves, not from vendors—that broadcast operators would push their own AI layer into the vendor tools and treat those tools as services the network calls. Adrian Koziol from Rogers described building a vendor-equivalent tool with a single developer in six months. Three different perspectives, same conclusion: the per-seat model is being hollowed out from both sides—customers building their own tools, and agents that don’t sit in chairs.

Sit with the implication. In that world, the user of a vendor tool isn’t a person at a seat. It’s an agent—running thousands of actions a day, often unattended, often in parallel with other agents. That breaks per-seat pricing in a way the industry hasn’t fully absorbed.

A vendor facing this has three choices. Cap the agent’s actions—which kills the value the agent was supposed to deliver. Price by usage openly—which gives up predictable revenue and exposes which workflows actually pay for themselves. Or hold per-seat and absorb the variable cost—which works until margin compresses, and the re-pricing conversation that follows is rarely a comfortable one.

Plenty of vendors are sitting in that third position today. It works while agent traffic is small. It stops working when it isn’t.

What’s actually landing at the CFO desk

For studios, broadcast networks, and any video business operating at production scale, the questions arriving at the finance table are different from the ones being asked three years ago. What’s our token budget across the toolchain next quarter? What’s our storage growth rate when AI multiplies the version count on every active show? What does an agentic workload look like across our pipeline, and where does it hit?

I don’t have clean answers to those questions yet. Neither does anyone I talked to last week. But finance teams burned once by cloud cost surprises don’t intend to repeat the experience with AI.

The infrastructure layer—network, compute, storage, and the orchestration that connects them—is where these conversations are landing. Three years ago the comparison was seat-vs-seat between vendor tools. Today it’s increasingly between architectures: how much you own, how much you rent, where the metering sits, and whether your infrastructure can absorb agentic workloads cleanly as they arrive.

The question that follows you home

The question from The Other Compression (April 12)—what happens to your costs when you iterate more?—is going to surface at every CFO review, every renewal cycle, and every board prep deck for the next two years. It sat underneath nearly every booth conversation I had last week.

But the more I think about it, the more I think the real question isn’t about what the next iteration costs. It’s about who in the pipeline absorbs the cost when iteration becomes the default, not the exception. Next time a vendor demos an AI feature, ask what happens to your invoice when you actually use it at scale. The pause before the answer is the answer.

Chuck Parker is CEO of Sohonet and serves on the boards of the HPA and GC-SC. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes (mostly) on the second and fourth Tuesdays of each month.

Two months later, the gap just closed by $600 million.

What Netflix actually bought

In March, Netflix acquired InterPositive, the AI filmmaking startup founded by Ben Affleck, in a deal Bloomberg reported could reach $600 million — potentially one of Netflix’s largest acquisitions ever. Affleck joined Netflix as a senior adviser alongside all of InterPositive’s staff.

What InterPositive built is worth understanding precisely, because it isn’t what most people assume when they hear “AI filmmaking tool.” The company’s patent, filed in 2024 and analysed in detail by Stephen Follows, describes a system that trains a custom AI model on a specific production’s own dailies. Not a general-purpose generative model. A per-project model that learns the cinematographic fingerprint of that particular show — its focal lengths, camera movements, lighting signatures, framing style, and editorial logic.

Once trained, the model handles tasks that currently consume significant time and budget: relighting shots, correcting continuity errors, removing stunt wires, reframing to fix composition problems, replacing or enhancing backgrounds. The system isn’t generating footage from prompts. It’s learning a production’s visual language and applying that understanding to solve concrete post-production problems.

The cost projections in the patent filing are aggressive. InterPositive’s internal documents, reported by Deadline, claimed 10–20% savings on below-the-line costs overall, 50% reduction in VFX costs, and 70% savings on background actors and stand-ins. On a specific budget example, they projected shaving $7 million off $32.1 million in below-the-line spend.

Those are projections, not audited results. But Netflix paid up to $600 million for them, which tells you something about how the buyer assessed the probability.

Process compression, not codec compression

Next week, NAB 2026 opens in Las Vegas. The show floor will be wall-to-wall AI messaging — V-Nova demonstrating LCEVC with 50 ecosystem partners, Appear showcasing low-latency hardware encoding, two dedicated AI Innovation Pavilions anchoring the floor. Real engineering advances that reduce bandwidth costs and improve delivery quality. But they’re solving for what happens after content enters the distribution pipeline. InterPositive sits upstream of all of it. It’s compressing the production process itself.

Traditional compression reduces the number of bits required to represent a finished frame. InterPositive’s model is a compressed representation of a production’s entire visual logic — not pixels but understanding: how this show looks, how its shots are constructed, what its visual rules are. And it uses that understanding to eliminate work that would otherwise require crew, equipment, and time.

Netflix is building the toolkit

The InterPositive acquisition doesn’t exist in isolation. On April 3, Netflix open-sourced VOID — Video Object and Interaction Deletion — a physics-aware model that removes objects from video while preserving scene plausibility. In user studies, it was preferred 64.8% of the time against competitors including Runway. It’s on GitHub under Apache 2.0.

Put InterPositive and VOID side by side and a strategy emerges. One is proprietary and acquired for hundreds of millions. The other is open-sourced for free. Both compress the production process. In the language of our Six Market Forces thesis, this is Generative-AI Cost Compression (Force 1) converging with Keeping Up With Netflix (Force 3). Netflix isn’t just using AI tools — it’s vertically integrating them, building an internal AI production capability that competitors will need years to replicate.

Running the numbers

So what does this mean in practice? Sohonet’s Screen Production Index tracks 127 Netflix productions over the last twelve months. By budget tier: 4 are X-Large ($100M+), 15 are Large ($35–100M), 41 are Medium ($10–35M), 48 are Small, and 19 are X-Small. That’s 60 productions in the Medium, Large, and X-Large tiers where InterPositive’s technology could materially impact below-the-line spend — titles like 3 Body Problem ($140M), Wednesday ($80M), Bridgerton ($40M), One Piece ($110M), and Outer Banks ($25M).

Using industry-standard assumptions — below-the-line costs represent approximately half of total production budget — and applying InterPositive’s own 10–20% BTL savings range from the Deadline-reported patent filing, the arithmetic is straightforward. Across those 60 M/L/XL Netflix productions, aggregate budget totals roughly $2.2 billion. Below-the-line spend: approximately $1.1 billion. A 10–20% reduction yields $108–216 million in potential annual savings for Netflix alone.

Those are projections built on claimed numbers, not audited results. Treat them as directional. But even the conservative end — over $100 million annually on a portfolio Netflix refreshes every year — raises a question that matters more than the dollar figure itself.

Where does the money go?

There are three possible destinations for AI-driven production savings, and the split between them will define whether this is a margin story or a volume story.

The first possibility is that savings get absorbed into future budget baselines — the most common pattern in production history. Once a cost reduction is demonstrated, the line item shrinks in the next budget cycle. Productions don’t get cheaper; they get squeezed.

The second possibility is that savings improve platform profitability directly. Netflix channels that $108–216 million into live sports rights, podcast deals, games, or simply takes the margin improvement. The production ecosystem sees no incremental volume.

The third possibility — and the one worth watching — is reinvestment into additional content. If even half of those savings flow back into new commissions, that’s $54–108 million in freed capital. At Medium-tier budgets (~$20M average), that funds 3 to 5 additional productions per year. At Small-tier budgets (~$5M average), it funds 11 to 22.

Here’s why the third scenario is more likely than it might appear: competition. The Six Market Forces thesis identified Keeping Up With Netflix as a distinct force because the streaming landscape is a volume race. In a zero-sum attention market, taking savings as pure margin while competitors use AI to produce more is a strategic risk Netflix is unlikely to accept. We saw this dynamic in 2023–2024: studios that resumed production fastest post-strike captured disproportionate share. In a content race, standing still is falling behind.

The real inflection point

Now extend the logic beyond Netflix. The SPI’s January 2026 report shows total industry production value of approximately $11.7 billion across all major studios. Netflix accounts for roughly 18% of that. The industry’s tier distribution skews more heavily toward Medium than Netflix’s portfolio does — 24% of all productions are Medium-tier, compared to Netflix’s 32%. Scale the same BTL assumptions across the industry’s approximately 295 M/L/XL productions, and the numbers shift substantially: industry-wide BTL savings range from roughly $500 million to $1 billion annually.

Here’s where it connects to the broader thesis. Sohonet’s Six Market Forces framework projects that pro-grade title counts will grow from roughly 1,960 today to 2,840 by 2030 — almost entirely concentrated in Medium and Small budget tiers, with X-Large and Large titles remaining flat. The M-tier alone is projected to grow by 179 titles over five years. If AI savings reinvestment funds even 13 to 26 additional M-tier productions per year at steady state — and adoption ramps gradually — AI cost compression alone could account for a quarter to a third of the projected M-tier title growth over the next five years. At the S-tier level, the numbers are even more pronounced.

That’s not a rounding error. That’s the mechanism by which Force 1 (AI Cost Compression) translates into the volume growth the thesis projected — concentrated in exactly the budget tiers where savings stretch furthest and where streamers are competing hardest.

What to watch for at NAB and beyond

NAB next week will surface dozens of AI tools across the production and distribution pipeline. The ones worth paying closest attention to aren’t the ones with the most impressive demos — they’re the ones that compress process, not just bitrates. That’s where the economics change.

The evidence isn’t just piling up anymore. It’s being acquired, integrated, and deployed. The question from February stands, updated: is your infrastructure, your workflow, and your business model ready for what happens when the savings start flowing back into the pipeline?

Chuck Parker is CEO of Sohonet. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

The tension surfaced immediately. Someone from the post-production side warned that costs are rising on both sides of the ledger — RAM, NVMe, and networking gear on-prem; storage, compute, and egress in the cloud. A studio engineer argued that none of it matters if the artist experience isn’t seamless — creatives shouldn’t have to know or care where their infrastructure lives. Fotter’s argument cut through both: the real promise of public cloud is elastic scale — the ability to spin up and spin down compute as a project demands it. The friction, he argued, is partly skeuomorphic. We treat the cloud like a separate computer rather than an extension of a workflow, one where specific components justify the premium of dynamic compute. But the friction is also economic: cloud costs are opaque, unpredictable, and hard to model against on-prem alternatives. And someone at the table asked the question that nobody could fully answer: is any of this even the highest-leverage cost problem in production, or are we optimising the wrong line item?

It was a room full of smart people wrestling with the economics of hybrid infrastructure — and nobody had a clean answer.

That’s because the question has changed. The industry spent a decade debating “cloud versus on-prem” as though it were a binary choice. That argument is over. The answer is “both.” But “both” has its own economics, and our industry hasn’t fully reckoned with them.

Where the industry actually sits

According to Flexera’s 2025 State of the Cloud report, enterprises now use an average of 2.4 public cloud providers, and 70% have adopted hybrid cloud strategies combining public and private infrastructure. This isn’t hybrid in the old sense of on-prem-plus-one-cloud. It’s multi-cloud-plus-on-prem, often with different providers handling different workflow stages. The landscape has fragmented well past the binary debate.

Meanwhile, 86% of CIOs report planning some form of cloud repatriation. That sounds like a retreat, but it isn’t. It’s a correction. Workloads are being re-sorted: bursty render and transcode stay in the cloud; persistent storage, review, and creative workstations are coming back to managed or owned infrastructure. The industry isn’t moving in one direction — it’s oscillating, and the oscillation itself has costs.

What “hybrid” means on the ground is something like this: dailies reviewed in London, VFX rendered in Montreal, conform in Los Angeles, deliverables pushed to a distributor’s cloud bucket in Virginia — all on the same production. Media workflows don’t flow “up and down” to the cloud the way traditional IT workloads do. They flow laterally — venue to facility, ingest to playout, artist to reviewer. Different stages of the workflow have fundamentally different requirements, and they belong in different places.

Three forces making the economics harder

The first is the widening gap between what compute costs to build and what it costs to rent. GPU price-performance has improved at roughly 29% per year since 2019, according to analyses of GPU benchmark-per-dollar trends. Yet hyperscaler pricing hasn’t tracked that curve. AWS on-demand compute pricing actually increased by approximately 23% between 2020 and 2023, and while selective cuts followed in mid–2025, the January 2026 GPU price hike clawed some of that back. The net effect: the underlying cost of compute keeps falling, but the price you pay on public cloud has been volatile — and in some periods, moving in the opposite direction.

This doesn’t make public cloud the wrong choice. It makes it a choice that needs to be made deliberately. The elasticity premium — the ability to burst up and tear down on demand — is entirely justified for episodic, high-burst workloads: a render farm that spins up for three weeks and disappears, a disaster-recovery failover that fires once a year, a transcode pipeline that scales with ingest volume. It makes far less sense for always-on workloads — your NAS, your creative workstations, your persistent review infrastructure. The discipline is knowing which workloads justify the premium and which don’t. Most organisations haven’t done that maths rigorously.

The second force is data gravity and the hidden egress tax. The same Wasabi survey found that 51% of media and entertainment cloud spend on object storage goes to fees — egress charges, API calls, retrieval costs — rather than actual storage capacity. That’s higher than the cross-industry average. More than half of the organisations surveyed reported IT or business delays caused directly by egress costs. Once assets land in a cloud region, the cost to move them out frequently exceeds the cost of storing them there in the first place. Productions don’t budget for this. They plan for storage and compute, then discover data movement on the invoice.

The third force is file-size inflation. 8K acquisition, ACES and HDR colour pipelines, higher frame rates, volumetric capture from LED stages — the raw data footprint per shot keeps compounding. A single minute of 8K camera-original footage in a compressed RAW codec like RED’s R3D runs roughly 60 gigabytes — and uncompressed, it approaches 300. Even at delivery-grade compression, HDR adds 25 to 50% to bitrates compared with SDR at the same resolution. This is the accelerant that makes the other two forces worse. The hyperscaler premium hits harder when there’s more data to process. Egress compounds when there’s more data to move. And AI demand for GPU and compute resources is simultaneously creating scarcity on the very hyperscaler capacity that was supposed to provide burst flexibility — media companies are now competing for the same silicon as the large language model operators.

The chokepoint nobody is watching

Doug Shapiro, in his forthcoming book Infinite Content (MIT Press, 2026), makes a compelling argument about price deflation in media. His thesis, in brief: when the cost of creating content approaches zero — as generative AI accelerates — value migrates to the scarce complements. Distribution, curation, and infrastructure.

Infrastructure is the one worth watching. Unlike distribution or curation, infrastructure capacity is physically constrained. GPUs have to be fabricated. Fibre has to be laid. Data centres need power and cooling. You cannot generate more infrastructure with a prompt. And yet the entire AI-accelerated production pipeline depends on it.

This is the squeeze. Content costs are deflating. The tools to create, iterate, and refine are getting cheaper and faster. But the infrastructure required to process, store, move, and review all of that content is facing pricing pressure from multiple directions — the structural gap in hyperscaler economics, the compounding effect of larger files, and AI-driven scarcity on burst compute. Production companies sit in the middle.

What this means heading into NAB

AI tools will accelerate iteration. That much is certain. There will be more shots, more versions, more review cycles per project than ever before. Each of those shots must be stored, moved, modified, colour-graded, reviewed, and delivered. That incremental iteration has a real cost — and if it’s priced on a consumption model, the economics of faster iteration actually work against you. The tool that made you faster also made you more expensive.

Understanding the cost components of shot creation by workflow stage — compute, storage, movement, review bandwidth — is becoming a core competency. Use public cloud where burst and elasticity justify the premium. Keep persistent workloads on infrastructure you own or lease at predictable rates. Find partners whose pricing model doesn’t penalise velocity.

That’s what the conversation at Jason Fotter’s table was really about. Not cloud versus on-prem. Not which hyperscaler wins. But understanding which workflow component belongs where, and why — and building the discipline to act on that understanding before the invoice arrives.

NAB 2026 will be wall-to-wall hybrid messaging. Every vendor on the floor will have a story about cloud-native workflows and AI-powered pipelines. The question worth asking each of them: what happens to my costs when I iterate faster?

Budget success on professional television and film projects increasingly depends on the answer.

Chuck Parker is CEO of Sohonet. The Screen Production Index and Six Market Forces Reshaping Hollywood report are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

Those are the macro numbers. They tell a story of gradual stabilisation and geographic diversification. If you stopped there, you’d walk away thinking the post-strike recovery is proceeding roughly as expected.

But inside the US, the state-level data tells a very different story. Within that 35% US share, New York claimed 31% of starts. California showed 23%. Georgia came in at 19%. Two years ago, California and Georgia between them accounted for nearly 60% of US production budget. Today, both are contracting — and for structurally different reasons.

I want to take the space this week to walk through what’s happening in each state, because I think Georgia and California are the two most important regional stories in the industry right now. Not because they’re growing — but because the forces driving their decline are different, and understanding those differences matters for anyone making decisions based on where production is heading.

Georgia: The Anchor Tenant Left

Georgia’s rise was one of the great success stories of the incentive era. A 20% transferable tax credit with a 10% logo uplift, combined with world-class facilities including Trilith Studios, Shadowbox Studios, Assembly Studios and Tyler Perry Studios, turned Atlanta into a genuine rival to Los Angeles. In 2022, Georgia recorded 84 productions worth $1.81 billion — 24.6% of all US production budget. Marvel tentpoles, Disney+ series, and major streaming originals were flowing through the state at scale.

The strikes hit Georgia disproportionately hard: a 67% drop in production count and an 89% drop in budget. The 2024 bounce looked encouraging — 63 productions, $1.0 billion — but it wasn’t sustained. By 2025, production count had fallen back to 44, and the state’s US market share sat at 11.3%. Mean budget per project dropped from $27.9 million to $13.4 million.

Source: Sohonet SPI Focus — Georgia, January 2026

The structural issue is the Marvel/Disney shift. In late 2024, Marvel Studios confirmed a rebalancing of its production footprint toward the UK, with several upcoming projects moving to Pinewood Studios. This reflects both the changing pound-dollar exchange rate and the UK’s 25-40% incentive structure. While Captain America: Brave New World filmed primarily in Georgia, future MCU tentpoles are increasingly likely to shoot internationally.

But the anchor-tenant problem is compounded by a broader set of competitive pressures. International incentive programmes — UK at 34% base, Ireland at 40%, Germany at 30% — have become structurally more attractive. Georgia’s own credit programme, while recently reformed with audit improvements and a new standalone post-production credit in 2026, still faces perception challenges around comparative labour costs, healthcare premiums on US talent, and a cultural pull that hasn’t kept pace with the infrastructure investment. The 2026 pipeline currently shows 11 confirmed projects worth $209 million — a fraction of peak activity.

In the language of our Six Market Forces thesis, Georgia is primarily a Tax-Incentive Migration story. The state built a production economy on the strength of its credit and the gravitational pull of a few major tenants. When those tenants recalculated — because currency moved, because competing credits improved, because streamers tightened budgets — the ecosystem proved less sticky than the infrastructure suggested.

California: Bigger Share of a Smaller Pie

California’s situation is deceptively different. On the surface, the state’s share of US production budget has actually risen — from roughly 33% in 2022 to approximately 43% in 2025. That looks like resilience. It isn’t. The total US production budget contracted over this period, and California’s absolute numbers tell the real story: 150 productions in 2022, down to 80 in 2025. Aggregate production budget fell from $2.44 billion to $1.93 billion.

Source: Sohonet SPI Focus — California, February 2026

California’s structural advantage is that it’s home to the major studio lots — Disney, Warner Bros., Universal, Paramount, Sony — and the concentration of above-the-line talent, creative infrastructure, and decades of established relationships creates a baseline of high-budget studio production that no incentive programme alone can replicate. Post-strike recovery has largely concentrated on this studio lot-based activity.

The problem is beneath the headline. The independent production market — historically a significant secondary layer of California’s ecosystem — has been markedly slower to recover. With fewer streaming commissions, tighter budgets, and California’s capped, over-subscribed incentive programme offering little relief, independent producers are increasingly sensitive to incentive differentials and directing activity to states with more aggressive credits. The January 2025 LA wildfires added further disruption to an already subdued environment.

And the streamers themselves are recalibrating. Netflix and Amazon, which don’t operate traditional California lots at the same scale, have re-examined their in-state production mix against competing incentive structures. When viewed as a proportion of overall spend, streaming platform concentration in California has declined, while traditional studio concentration has held more steadily. The recovery, such as it is, belongs to the majors — not the broader ecosystem.

If Georgia is a Tax-Incentive Migration story, California is more a convergence of Linear-TV Cashflow Decline and the Keeping Up With Netflix dynamic. Studios have less free cash as linear revenue erodes, so they’re more ruthless about optimising where they spend. Streamers are investing at 4-6% CAGRs but pushing toward lower-budget, higher-volume slates where cost efficiency matters more. California remains the administrative and creative home of Hollywood — but the economic gravity that kept production volume high is weakening underneath.

Why the Regional View Matters

The national SPI numbers are designed to give you direction. They do that well. But the gap between the national trend and the state-level reality is unusually wide right now, and for anyone whose business depends on regional production activity — whether you’re a facility, a crew base, a services provider, or a local economy — the aggregate doesn’t tell you enough.

This is part of why we’ve started publishing regional SPI Focus reports. The Georgia report came out in January; California in February. They go deeper into the studio-by-studio breakdown, the incentive dynamics, and the pipeline outlook for each state. If the questions I’ve raised here feel relevant to your planning, the detail is there.

The broader point is this: the post-strike recovery is real, but it’s a redistribution, not a restoration. The states and regions capturing share — the UK, New York, New Jersey, the emerging international hubs — are benefiting from structural tailwinds. The states losing share are facing forces that won’t reverse on their own. The data is telling a story, and it’s worth reading carefully.

Chuck Parker is CEO of Sohonet. The Screen Production Index and SPI Focus regional reports are available at sohonet.com.

But I keep coming back to one question from the morning keynote. Doug Shapiro — the media analyst whose Substack and newsletter have become essential reading for anyone tracking entertainment economics — framed the entire day with one of the primary concepts of his 12^th Chapter of “Infinite Content” (draft) with a single organizing question: What becomes more valuable when content approaches free?

That question is worth sitting with, because it reframes almost everything our industry is debating right now.

The Second Symbolic Layer

Shapiro’s thesis is structural, not speculative. He argues that the last fifty years were a disruption of distribution — digitisation created a universal symbolic layer (bits) that dissolved the boundaries between media formats, made distribution functionally free, and set off the competitive dynamics we’re still living through: stagnant total viewing time, the weakening of intermediaries, and the fragmentation of attention across an explosion of content.

Generative AI, in his framing, is a second symbolic layer — one that dissolves the boundaries of creation the way digitisation dissolved the boundaries of distribution. The biggest change isn’t aesthetic or technical. It’s economic. Content creation goes from scarce and expensive to abundant and cheap.

Shapiro walked through the math. YouTube already uploads roughly 300 million hours of content annually — the equivalent of 20,000 Hollywoods. If even 0.01% to 0.1% of that output becomes — in Shapiro’s phrase — “not comparable but competitive for time,” the supply shock is enormous. And the definition of “quality” is already broadening. Our industry tends to define quality through production value — colour bit depth, high-end audio, the precision of VFX shots. Consumers haven’t stopped caring about those things. But they’re also showing evidence of valuing authenticity, relatability, relevance to their subcommunity, and emotional resonance in ways that don’t map neatly onto our definition. Growth in music is being driven by ambient uses. Growth in gaming is being driven by blocky worlds on Roblox while AAA studios spend hundreds of millions on cinematic graphics. YouTube is now the largest streaming option on the television, and social formats are competing directly for the same viewing hours. What matters is that all of these compete for the same fixed pool of attention.

Shapiro used the AI fan film “Beggar’s Canyon” as an illustration. The audience response wasn’t people ignoring the quality — it was people feeling something. Comments weren’t technical nitpicks; they were emotional: the film felt like the Star Wars they remembered. That’s not consumers abandoning production value. It’s consumer sentiment expanding what counts — and our industry needs to pay close attention to where that sentiment heads. As Shapiro put it, the correlation between success and quality is “probably going to break down.” Not because quality stops mattering, but because in a world of infinite supply, so much good content will exist that success increasingly depends on discovery, timing, and resonance rather than spend alone.

Shapiro also argued that the killer app for generative media is unlikely to be cheaper versions of what already exists — it’s more likely to be entirely new experiences that weren’t economically viable before. That’s a point worth holding onto, because most industry debate is still anchored on whether AI can replicate existing formats. The bigger disruption may come from formats we haven’t categorised yet.

Where Value Migrates

So: what becomes more valuable when content approaches free? Shapiro’s answer is that value flows to four places: challengers who exploit the new economics, consumers who benefit from abundance, complements that become more valuable when a paired input gets cheaper, and choke points that control scarce bottlenecks. For the production ecosystem, complements and choke points are the most actionable of the four.

Complements are the products and services that get more valuable when another input gets cheaper. When crop yields increased, value migrated to distribution, processing, and branding. When digital photography made images free, value migrated to editing, storage, and social networks. In media, when recorded music deflated, value migrated to live concerts. When games went free-to-play, monetisation shifted to skins, emotes, and community marketplaces. Content becomes top of funnel, not the product.

Choke points are the scarce capabilities that control value flow — discovery, measurement, monetisation, trust, provenance, first-party data. Shapiro pointed to search as the canonical example: it started as a complement when information exploded online, then became a choke point when Google consolidated it.

For the production and post-production ecosystem, this framing lands in a specific place. If content is abundant and attention is scarce, then the infrastructure between creation and consumption — the pipeline that ensures quality, manages complexity, enables collaboration, and delivers finished work — is a complement that becomes more valuable, not less. And in Shapiro’s own framing, some of that infrastructure — professional validation, systems integration, workflow orchestration — appears on the chokepoints slide as well. The distinction matters: complements create value when a paired input gets cheaper; chokepoints capture value by controlling scarce bottlenecks. Production infrastructure can be either, depending on how consolidated and hard to route around it becomes.

The Rest of the Day Through That Lens

What follows is my synthesis, not Shapiro’s. He delivered the macro thesis in the morning and the rest of the panels covered their own ground. But watching them through the lens of his framework, I kept seeing operational echoes of the dynamics he described — competition intensifying, disintermediation accelerating, and the question of who owns discovery and curation becoming more urgent.

The Integrated Workflows panel, featuring Autodesk/Wonder Dynamics, Blumhouse, and Topaz Labs, described a persistent split between traditional and AI pipelines. Every AI-assisted production is a “snowflake” — a bespoke workflow that may need to be re-architected mid-project as models update weekly. Their takeaway was blunt: don’t experiment with the non-experimental parts. Naming conventions, versioning, storage, approvals, QC — the operational backbone of production — becomes more important, not less, when the creative tools are changing underneath you.

The Generative Production panel reinforced this from a different angle. One animator described compressing a five-minute animation from a traditional one-to-two-year timeline down to 28 days. That’s a staggering acceleration. But the panel was equally clear about what didn’t get easier: character consistency, motion logic, maintaining coherent visual intent across scenes. The consensus was plain — AI doesn’t grant talent. In the hands of great artists it can enable greatness. In average hands it produces average work.

The War for Attention panel, with Sony, Warner Bros. Discovery, and Cineverse, landed on the discovery side of Shapiro’s equation. One panelist argued that metadata quality is the structural bottleneck in streaming — that search and discovery are broken because the data layer underneath them is shallow and fragmented. Another described using AI to scale UGC analysis across tens of thousands of posts for major franchises, extracting marketing playbooks that would have taken months manually. Both were describing complements — services that get more valuable as content volume rises.

And the closing Computable Studio panel — featuring WBD, BBC Studios, Google, Microsoft, Eleven Labs, and Asteria — converged on a prediction that reinforces several parts of our Six Market Forces thesis: content volume will explode, but the organizations that win will be the ones that treat data quality, workflow orchestration, and curation as first-class strategic capabilities. Multiple speakers flagged that compute and AI costs will become explicit budget line items as vendor subsidies fade, making operational efficiency and pipeline discipline critical.

What It Means for the Production Ecosystem

Shapiro’s thesis operates at the level of media economics. But for the people who actually make content — the producers, post houses, VFX facilities, and infrastructure providers — the implication is concrete.

There is no world in which a prompt generates a finished production. Even at dramatically lower price points, even with tools that compress timelines by 90%, the volume Shapiro projects still implies pipelines, workflows, review cycles, asset management, and quality control. The tools change. The need for the connective tissue between those tools does not. If anything, it intensifies — because more titles, at faster speeds, across more distributed teams, with tools that update weekly, is a harder coordination problem than what came before it.

In our Six Market Forces modelling, we project approximately 880 additional professional-grade titles by 2030, concentrated in small and medium budgets. Shapiro’s framework supports the directional logic — when creation costs compress, supply expands — though open questions remain, including whether AI savings fund more volume or simply shift to marketing, and whether audience growth tracks titles or stays hit-driven. What’s clear either way is that the broader production and post-production ecosystem — the workflows, toolchains, and operational infrastructure the industry relies on — must adapt.

The question I posed at the end of the HPA blog two weeks ago still holds. The exact outcome is up for debate — how much volume, who captures what, which formats emerge. But the direction of travel is not. This summit sharpened that. It’s not just about speed or cost. It’s about what becomes valuable when content approaches free — and that requires innovating your workflows, your infrastructure, and your business model now, not once the answers are all in.

Chuck Parker is CEO of Sohonet. The Six Market Forces Reshaping Hollywood report and Screen Production Index are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month.

In Monday’s blog I covered a cluster of marketplace announcements that suggested our AI cost compression thesis may be playing out faster than projected. By Thursday afternoon, the Retreat had reinforced that read, not just on cost, but on the speed at which AI is embedding into actual production infrastructure.

Of roughly 35 sessions across four days, more than 20 featured AI explicitly in the title or as a primary subject. Tuesday’s Supersession - “WTF Is Going On in the Media Industry?” - was almost entirely devoted to AI. Even panels nominally about standards, preservation, or sports media circled back to AI in their discussions of metadata, embeddings, and automated workflows. This wasn’t a track within the program. AI was the program.

But the real story isn’t the count. It’s the spectrum. A year ago, the AI conversation at industry events clustered around text-to-video generators and hypothetical timelines. This year, it spanned from Robert Tercek projecting dramatic per-minute cost reductions in the near future, all the way to Wednesday afternoon’s demonstration of multi-agent systems that route director notes into department-specific timecoded instructions. The distance between visionary forecast and operational tooling is where the real action is, and the Retreat covered that entire range.

The intellectual anchor of the week was Ed Ulbrich’s Tuesday Supersession keynote. Ulbrich - now Head of Strategic Growth at Moonvalley, the licensed-data AI video model company backed by $154 million - opened with a line that set the tone: the future does not fit in the containers of the past. His core argument was that this is not “AI-first” time but “judgment-first” time. Creative and business judgment must precede tool selection. You don’t get extra credit for the tool; what matters is the piece.

What made his talk land wasn’t the framing, it was the specificity underneath it. He described a dual conversation happening across the industry: publicly, caution and polite talking points; privately, active pilots, real experimentation, and people making things they like. And then the line that stayed with me through the rest of the week: the nightmare is not our extinction, it’s mediocrity scaling faster than what we do. That’s irrelevance. Without judgment and governance, the default future isn’t a renaissance, it’s slop at mass scale. Ulbrich’s four ethical pillars - consent, compensation, credit, and commercial safety - framed as operational requirements, not aspirational goals, became a recurring reference point for every session that followed.

The practitioners who followed him on Tuesday were just as direct. Nik, an Emmy-nominated filmmaker and one of OpenAI’s original “Sora Seven” was blunt: flashy demos hide real production issues. Consistency and directability are the major obstacles. His “Critterz” project - nine months of production instead of three years, a budget roughly one-seventh of a comparable Pixar release - is the most concrete example I’ve seen of the Generative-AI Cost Compression force playing out in real time. Jason Zada of Secret Level reinforced the same point: real progress is measured by filmmaker control, not brute-force rerolling.

Not everything was optimistic. Wednesday’s compositing session named something I’ve been hearing privately for months: faster turnarounds plus lower rates plus smaller teams equals more multitasking and busier artists, not freed-up creative time. This is the efficiency trap I flagged in the Market Forces report, that cost compression can be captured by the buyer rather than the maker. Studios demanding more for less, rather than artists doing better work with better tools. The compositors on that panel were candid: traditional VFX skills remain non-negotiable for refining AI outputs, but the role is transforming from late-stage assembly into a creative control hub. The “cross-disciplinary ninja” who blends skills across traditionally separate departments is now the hiring profile, and those people are hard to find. It’s the same talent question I raised in January and the Retreat didn’t resolve it.

Governance ran through nearly every session as an operational reality, not a policy discussion. Maintain detailed authorship logs. Disclose proactively. Document everything. Every successful AI-assisted production case study presented during the week, Nik’s branded films, Secret Level’s Liquid Engine work, the ETC’s “Tale of Two AI Productions” had one thing in common: visible, documented human authorship at every stage. Not as a defensive posture, but as a creative differentiator. The HPA Awards nominees this year include AI tools across five distinct production lifecycle categories, from Yamdu’s AI Script Breakdown in pre-production to Foundry’s Nuke Stage in VFX. The tool landscape has matured from novelty to nomination-worthy in roughly 18 months.

Here’s where I net out. The February SPI data showed the Q1 recovery converting from starts into actual spending, the leading indicators are moving in the right direction. The Retreat compressed four days of industry conversation into the same signal. When the agenda shifts from “will this work?” to “how do we govern it, staff for it, and build infrastructure around it?”, the question becomes when that shows up in the production data, not whether.

My best guess: we will see the early indicators in the back half of 2026, concentrated in the Small and Medium tiers. The question we are all asking ourselves is whether the efficiency gains accrue to the makers or stay captured by the buyers. If the average budget declines but there is not incremental volume, the buyers are capturing the value. If the volume of productions in those lower budget tiers increases, then the created value is being reinvested.

———

Chuck Parker is CEO of Sohonet. The Six Market Forces Reshaping Hollywood report and Screen Production Index are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month. This post is the second of two special editions from the 2026 HPA Tech Retreat.

In Sohonet’s Six Market Forces Reshaping Hollywood report, we identified Generative-AI Cost Compression as the first of six forces that will define content production through 2030. The thesis is straightforward: AI is compressing the unit cost per shot by enabling smaller teams to work faster across assembly edits, rotoscoping, layout, and iteration. Unlike earlier tech cycles where savings were absorbed by extra work, this shift allows creators to try more ideas, move through cycles faster, and achieve better results, while stretching lower production budgets further. The report projected that a 10% annual cost drop in small and medium budgets would lead to roughly 10% more titles per year in those tiers.

That was a forward-looking projection. But in just the last two weeks, the marketplace has delivered a cluster of announcements that suggest the projection may be conservative.

Amazon MGM Studios Goes to Beta

On February 4th, TechCrunch and Reuters reported that Amazon MGM Studios will begin a closed beta program in March, inviting external industry partners to test proprietary AI production tools developed by its internal AI Studio. The unit, launched last summer under Albert Cheng, has been collaborating with producers like Robert Stromberg (Maleficent), Kunal Nayyar (The Big Bang Theory), and Colin Brady (former Pixar animator) to bridge what Cheng described as the “last mile” between existing AI offerings and what directors actually need on set.

This isn’t theoretical research. Amazon’s House of David series already used 350 AI-generated shots in its second season, where director Jon Erwin combined AI with live-action footage for battle scenes. The tools focus on character consistency across shots, pre- and post-production support, and integration with industry-standard creative applications. Amazon is working with multiple large language model providers and expects to share initial outcomes by May.

The framing from Cheng is worth noting: “It is expensive to produce, and indeed, it is difficult to produce more, and it is difficult to take a lot of risk.” That’s the economics of content creation in 2026, stated plainly. And it maps directly to the Market Forces thesis — when production costs constrain volume, cost compression tools don’t just save money; they unlock titles that otherwise wouldn’t get greenlit.

Amazon isn’t operating in a vacuum here. Netflix disclosed last July that its series The Eternaut used generative AI for a building collapse scene that was completed 10x faster than traditional VFX workflows. Co-CEO Ted Sarandos said the cost of producing it through conventional means simply wouldn’t have been feasible for a show with that budget. Netflix’s internal Eyeline Studios VFX unit drove that work, and Season 2 — renewed for late 2026 — is expected to lean harder into AI for pre-visualization, shot planning, and select VFX sequences. Disney went a different direction entirely with a $1 billion OpenAI deal in December 2025 focused on consumer-facing AI content creation using Disney IP on Disney+.

Three major studios, three distinct AI strategies: Amazon building vertically integrated production tools backed by AWS compute, Netflix embedding AI into its in-house VFX pipeline, and Disney licensing IP for consumer-facing AI generation. Each creates different infrastructure demands, but all three compress costs or extend reach in ways that weren’t possible two years ago.

A $250 Million Bet on AI-Native Production

Four days later, Variety reported that Filmology Labs, a $250 million studio complex, will open in Paterson, New Jersey. Founded by Alan Mruvka, co-creator of E! Entertainment Television, the 250,000-square-foot facility is purpose-built for a different kind of content: vertical micro-dramas, digital series, podcasts, and AI-driven media formats.

What’s notable isn’t just the dollar figure — it’s the design philosophy. Filmology Labs will feature 21 production sound stages with pre-built, fully lit, permanent standing sets optimized for speed and efficiency. The campus includes dedicated AI studios, an LED Volume Wall stage, and takes advantage of New Jersey’s production tax credits. The explicit goal is to eliminate the time and cost typically associated with set design, construction, and lighting — allowing content to be produced quickly and economically.

This is the physical infrastructure of creator-economy aggregation meeting AI cost compression. It’s the kind of facility I anticipated in the Market Forces report: new aggregation points where agency-backed and creator-driven micro-studios will produce at scales that didn’t pencil out before AI entered the workflow.

Cross-Media AI Platforms Emerge

On February 11th, Variety reported that iwot Studios (owner of The Wheel of Time IP) and VFX company Framestore are launching an AI-enabled platform designed to unify asset management across film, television, video games, animation, advertising, immersive experiences, and user-generated content — all from a single governance layer. The first IP to use the platform will be The Wheel of Time, which is simultaneously being developed as an animated film, a live-action movie, and a AAA video game.

This is a different flavor of AI application — not generating shots, but managing the complexity that comes from exploiting IP across multiple formats simultaneously. It’s a direct response to the operational challenge I flagged in the report: more iterations increase complexity, and studios need flexible, scalable production infrastructure to manage it. The fact that a VFX house like Framestore is co-building this platform signals where the traditional post-production sector sees its future value.

Meanwhile, In a Cinema Near You

And then there’s Critterz, the Vertigo Films animated feature produced using OpenAI tools, with a Cannes preview announced. The production timeline: nine months instead of the three years typical for a conventional animated feature. The estimated budget: roughly €25 million, compared to €170 million for a comparable Pixar release. The script was written by the Paddington writing team, but the images were entirely generated from AI models. Whatever you think about the creative merits, the economics are unambiguous.

The HPA Tech Retreat as a Leading Indicator

These announcements aren’t happening in isolation. The HPA Tech Retreat agenda itself reflects the same convergence. Consider what’s on the program this week:

The Tuesday Supersession features Ed Ulbrich — now Head of Strategic Growth at Moonvalley, the licensed-data AI video model company backed by $154 million from investors including Khosla Ventures, General Catalyst, and CAA — speaking on the artistry and ethics of generative tools. Jessie Hughes from Leonardo.AI will share what she described as “this month’s leading cinematic video workflows” and how studios are adapting in real time. Hughes has trained the US Television Academy and Hollywood studios on these tools behind closed doors for years; the fact that this is now a headline conversation tells you how fast the Overton window has moved.

A Creator Economy panel pairs Emmy-winner Bernie Su with Chad Nelson from OpenAI and Pojo Riegert from CrunchLabs. A business panel moderated by Seth Hallen features executives discussing how to respond to — and benefit from — disruption across the media supply chain. The midweek roundtables cover AI-assisted editing and cloud-era media pipelines.

Even the HPA Awards Innovation & Technology nominees (to be presented Wednesday) read like a map of where AI is embedding across the production lifecycle: Bria AI’s GenAI Attribution Technology and Yamdu’s AI Script Breakdown in pre-production, Adobe’s Generative Extend in Premiere and Flawless’s TrueSync visual dubbing system in post-production, and Foundry’s Nuke Stage alongside Nerfstudio and ILM’s collaboration in VFX and virtual production.

It’s worth stepping back and noting the breadth here. A year ago, the AI-in-production conversation was largely about text-to-video generators and hypothetical workflows. Today, the industry has AI tools nominated for awards across five distinct production lifecycle categories — from script breakdown to distribution. The tool landscape has matured from novelty to nomination-worthy in roughly 18 months. Companies like Moonvalley, whose Head of Strategic Growth Ed Ulbrich is keynoting the Supersession, have raised over $150 million to build licensed-data AI video models specifically for professional production use. The supply side of AI production tools is scaling to meet what appears to be accelerating demand.

What the Data Doesn’t Show Yet — and When it Might

Here’s my honest assessment: Sohonet’s Screen Production Index doesn’t yet show a definitive volume shift driven by AI cost compression. The SPI is a lagging indicator by design — it measures what’s actually being produced, not what’s been announced.

But leading indicators don’t stay leading forever. When Amazon is beta-testing AI production tools with industry partners, when a $250 million studio complex is being built around AI-native workflows, when VFX houses are co-developing AI asset management platforms, and when the industry’s premier technology retreat devotes its flagship session to asking “WTF is going on?” — the gap between announcement and data narrows - and we should expect those results to begin showing up in the production data in the back half of 2026 (e.g. higher counts in lower budget tiers / smaller t-shirt sizes).

The Six Market Forces thesis projected approximately 880 additional professional-grade titles by 2030, concentrated in the Small ($1-10M) and Medium ($10-35M) budget tiers, with growth coming from AI cost improvement (freeing up investment capital for more projects), streamer volume pressure, and creator-economy professionalization. The announcements of the last two weeks indicate the industry leaders are chasing a similar trajectory.

The question for production infrastructure providers, post-production houses, VFX studios, and everyone in between isn’t whether efforts to utilize AI to go faster and stretch budgets further is real. The evidence is piling up. The question is whether your workflows, your infrastructure, and your business model are ready for the change that’s coming.

I’ll be watching the data closely. Stay tuned.

———

Chuck Parker is CEO of Sohonet. The Six Market Forces Reshaping Hollywood report and Screen Production Index are available at sohonet.com.

Storytellers & Silicon publishes on the second and fourth Tuesdays of each month. This special edition coincides with the 2026 HPA Tech Retreat.

One month in, the answer is yes. The Q1 2026 production count is now projected at +52% on Q1 2025, up from +32% last month. The smoothed budget is trending upward — up 9% month-on-month, with the pipeline entering production now exceeding the value of productions wrapping up. The recovery is converting from starts into actual spending. For those keeping score against the peak: January was up 24% on January 2025 but still down 34% on January 2022. We’re recovering. We’re not back.

But the most interesting signal in this month’s Sohonet Screen Production Index isn’t the headline number. It’s what’s happening at Warner Bros.

Before we get there, a few other things worth flagging. The UK’s growing share of production, which I called “sticky” last month, isn’t just holding — it’s accelerating. The UK captured 27% of new production starts in January, up from 22% across 2025, and smoothed production budget is up 93% year-on-year. The US, by contrast, is down 29%. Within the shrinking US market, California is a notable exception: its share of US production budget has grown from 33% in 2022 to 43% in 2025, with smoothed data nearly back to January 2022 levels. And Apple continues to execute at a striking pace — $440M committed in January alone, with 22 productions confirmed for 2026 against a full-year 2025 total of 26. Each of these deserves its own analysis, and we’ll return to them. But for now, Warner.

The Paradox

The consensus view on Warner Bros. Discovery is straightforward: the prospective acquisition — whether by Netflix, Paramount, or another buyer — creates uncertainty. Uncertainty delays greenlight decisions. Production slows. The studio enters a holding pattern until the deal resolves.

The data tells a different story.

Warner’s production count has been essentially flat from 2023 through 2025, having dropped significantly from 2022 levels. On the surface, that looks like the stasis narrative playing out. But underneath, total budget commitment roughly doubled in 2025 versus 2024. Warner is making the same number of bets, but each bet is getting substantially larger.

The forward pipeline makes this even more pronounced. Of the 16 ‘XL’ productions (budgets exceeding $100M) currently visible in the 2026 pipeline, five belong to Warner: Superman: Man of Tomorrow, The Batman 2, Minecraft 2, Paris Paramount, and Ocean’s 14. That’s nearly a third of the industry’s biggest upcoming projects, from a studio that by conventional wisdom should be in wait-and-see mode.

This isn’t a studio hedging. It’s a studio loading up on franchise tentpoles — proven properties with global audiences and built-in demand. Whatever the strategic rationale, the observable output is clear: Warner enters 2026 with a production footprint that comprehensively bucks the narrative of a studio in stasis.

The Concentration Question

Warner’s approach — fewer productions, much bigger budgets, concentrated in franchise IP — creates a different kind of market footprint than, say, Netflix’s high-volume model.

Five XL productions with confirmed timelines is a significant block of high-value work moving through the ecosystem. Tentpole franchise productions tend to hold their schedules; there’s too much marketing, distribution, and franchise planning layered on top for casual slippage. In that sense, Warner’s pipeline carries more certainty than its corporate situation might suggest.

But it also concentrates exposure. If an acquisition closes and the new owner restructures the slate, the impact isn’t distributed across dozens of mid-tier projects — it lands on a handful of very large ones. That’s a meaningfully different risk profile, and it’s one that ripples through every stage operator, VFX house, and post-production facility with Warner work in their pipeline.

What to Watch

The question for the next few SPI updates is straightforward: does Warner’s pipeline convert on schedule?

If those five XL productions shoot on time, it tells us that M&A uncertainty isn’t translating into production paralysis. If they start slipping, the paradox resolves in the other direction — and the downstream impact would be material given the budget concentration.

The broader market is healing. But the most telling data points right now aren’t the averages — they’re the outliers that defy the expected pattern. Warner is the biggest one.

Meanwhile, they’re pursuing an $82.7 billion acquisition of Warner Bros. Discovery’s studio and streaming assets, fending off a hostile bid from Paramount, and planning to spend nearly $20 billion on content this year—up 10% from 2025.

The question for those of us in the production ecosystem isn’t whether Netflix is winning. That’s been settled. The question is: what kind of company are they becoming, and what does that mean for how content actually gets made?

The Chorus of Observers

Evan Shapiro put it bluntly in his 2026 Agenda: “By summer, streaming will push over 50% of all television viewing for the first time, FAST will have put Netflix in its rearview, and YouTube will surpass all of US Broadcast, combined.” His point isn’t just about market share—it’s about the pace of change and where audience attention is flowing.

Doug Shapiro, writing in Fortune on the WBD deal, captured something more structural: the deal signals that Netflix believes the media landscape has “fundamentally shifted.” Traditional media, he argues, cannot win the time game against an infinite number of creators—many of whom are now being empowered by new tools including AI. That’s worth watching, though the implications for traditional production aren’t as immediate as some headlines suggest.

Jim Louderback offered perhaps the most useful framing: Netflix is “buying dinosaur armor.” Their model is “expensive but sturdy,” while YouTube’s is “cheap but jittery.” Two different philosophies of content economics coexisting in the same market. For producers and post houses, the question is which clients you’re serving—and whether you can serve both.

Lucas Shaw, who’s been inside the WBD deal reporting for Bloomberg, observed that Netflix has abandoned its “build, not buy” heritage. That’s a significant cultural shift. And Matt Belloni, tracking the regulatory and competitive dynamics at Puck, noted that the deal would “decisively end the streaming wars”—which means fewer buyers, but also potentially more stability in how the remaining majors operate.

What the Earnings Confirm

Five of the six forces we track got validation last week. Here’s how to read each one:

Linear-TV Cashflow Decline. This is why the WBD deal exists. Warner Bros. Discovery’s cable EBITDA dropped 21% year-over-year. The linear TV business that once funded content production is in structural decline, and that’s pushing studios to redeploy capital toward streaming originals or seek consolidation. Netflix isn’t buying WBD because Warner is thriving—they’re buying because linear decline has repriced the asset. For producers, the marker to watch: how quickly does your buyer base shift from networks to streamers?

Keeping Up With Netflix. This force is about the competitive pressure Netflix exerts on everyone else. The Q4 numbers confirm that pressure is only intensifying—$51 billion revenue guidance, 31.5% margins, $20 billion content spend. Rivals must match that volume to retain subscribers. The WBD deal would consolidate Netflix’s lead and leave Disney as the only comparable content machine. For facilities, the marker to watch: are you positioned with buyers who can sustain this level of spend?

Hybrid-Cloud Economics. Most post houses are already operating in this world. What’s notable is Netflix signaling that their own production and post workflows remain hybrid—not all-cloud. The takeaway is about cost-to-serve. Cloud has advantages for burst capacity and remote collaboration, but cost at scale favors on-prem. For post houses operating on thinner margins than the content producers themselves, managing that mix remains critical. The marker to watch: are your infrastructure costs tracking with revenue, or diverging?

Tax-Incentive Migration. Netflix mentioned UK investment specifically in the context of their infrastructure buildout. The incentive landscape continues to shape where productions land. For facilities, the marker to watch: is your region’s incentive structure keeping you competitive—and are you positioned to capture work when it arrives?

Creator-Economy Aggregation. Got its clearest signal yet. Netflix is now licensing video podcasts from iHeartMedia, Spotify, Barstool, and The Ringer. Pete Davidson and Michael Irvin have Netflix podcast shows. This is exactly what the force describes: creator content being aggregated into professional distribution. For traditional post and VFX facilities, this represents opportunity—Mr Beast, Ms. Rachel, Mark Rober and other creator-economy productions are increasingly sophisticated, and they need professional post services. The marker to watch: are you set up to serve clients who operate on faster cycles and different budget structures?

Where the Uncertainty Lives

Here’s where it gets interesting—and where I’m still sorting through the data.

The sixth force, Gen-AI cost compression, is the one that should be keeping production executives up at night. Netflix spent $18 billion on programming in 2025 and plans to spend nearly $20 billion in 2026. That’s a 10% increase. On the surface, that looks like good news for content creators.

But look at where the money is going. Live events. Sports rights (NFL, WWE). Podcasts. Licensing deals with Sony, Universal, and Paramount. A vertical video feed designed to capture attention on mobile.

What’s not clear from the data is whether any of this increase is flowing to scripted content. Netflix doesn’t give specifics about the volume or budget of those productions. It’s possible that scripted volume is actually increasing because of lower cost-per-production economics. It’s also possible that the incremental spend is going to live events, sports, and podcasts—and as a consequence, not helpful to the classic film and TV production industry. We simply don’t have enough visibility to say which.

Sarandos has been careful on AI. He said at Q3 earnings that AI tools are “interesting” but won’t replace human storytelling. He’s threading a needle—acknowledging the technology without signaling that it’s about to disrupt the cost structure.

But the math tells a different story. If you’re expanding total content spend while diversifying into formats (podcasts, live events, games) that don’t require traditional post-production workflows... you’re quietly reallocating capital away from the ecosystem that built you.

The question I keep asking: is the scripted content machine expanding, or is it being held flat while new formats absorb the growth?

Doug Shapiro’s observation about “infinite creators empowered by AI” points to the tension. Netflix doesn’t need generative AI to make Stranger Things. But they might need it—or something like it—to fill a vertical video feed with clips, podcasts, and micro-content at scale.

What This Means for Production

For those of us in the business of supporting production workflows, the next 18-24 months are going to require serious attention to where the work actually goes.

Large and extra-large productions aren’t going away. But they’re not growing either. The volume growth is in smaller and mid-sized productions—and increasingly in formats that don’t look like traditional episodic television.

If you’re a post house built around high-end scripted drama, the Netflix of 2026 is a different buyer than the Netflix of 2022. Their content mix is expanding into sports, live events, podcasts, and mobile-first formats. The workflow requirements are changing. The technology stack is changing. The economics are changing.

That’s not a crisis. It’s an opportunity—if you’re paying attention to where the production landscape is actually headed.

I’ll be tracking this as the WBD deal progresses and the 2026 content slate takes shape. If you have a different read on the data, I’d love to hear it.

Chuck Parker is CEO of Sohonet and publishes Storytellers & Silicon twice monthly. The full Six Forces report is available on Sohonet.com.

The shape of production - where it happens, what budgets dominate, who’s doing it, and how much money is actually in-flight at any given time - has shifted. And those shifts look sticky.

The Q1 Signal Is Real - Just Don’t Over-Read It

The headline: Q1 2026 is forecast to be up 32% on Q1 2025, with 210 confirmed projects in the pipeline. The forecast has remained within 4% of last month’s report, so this isn’t a single-month whipsaw.

But context matters. That +32% is measured against a dismal Q1 2025, the post-strike hangover. Against Q1 2022, we’re still down 17%. And if you look at smoothed production activity, a measure that spreads each production’s budget across its active period to show how much spending is actually flowing through the system, 2025 was down 20% versus 2024 and down 38% versus 2022.

Plan for a busier Q1 than last year, but don’t budget like it’s 2022.

The Geography Has Moved

If you only take one “so what” away for strategy and operating plans, make it this: where production happens is changing.

From 2022 to 2025, the US share of new production starts dropped from 43% to 34%. Over the same period, the UK’s share grew from 15% to 22%. Canada held steady at approximately 25%.

This isn’t temporary displacement from strikes or one-off project decisions. The underlying economics, tax incentives, infrastructure maturity, labor costs, currency dynamics, have made the UK structurally more attractive for production. Like the Fed describing inflation, these are the market forces that keep shifts sticky: they don’t snap back when volume recovers.

If your planning assumptions still treat the US as the default center of gravity, you’re going to mis-allocate attention and capacity.

Three Studios, Three Strategies

The studio data reveals distinct strategic bets emerging from the contraction.

Netflix: Reach and catalog breadth. In a market where overall production count dropped 14% year-over-year, Netflix grew its production count by 10%. Their market share has expanded from 13% in 2022 to 18% in 2025. With $2.2B spent across 131 productions, their average budget sits around $17M. Netflix is optimizing for serving many audiences, many cultures, many niches.

Disney: Premium IP that travels. Disney remains the largest spender at $2.8B, but across only 93 productions, an average budget around $30M, nearly double Netflix’s. Their share contracted from 27% to 23%. Disney is making fewer bets, but each one is designed to work globally on brand strength and established IP.

Apple: Swinging for the fences. Apple produced just 26 titles in 2025 at $729M total (~$28M average), but their upcoming pipeline tells the real story. Of the 56 L/XL projects visible in the forward pipeline, Apple has 10, the same number as Disney, despite producing roughly one-quarter the volume. Nearly 4 in 10 Apple productions are premium-tier. Netflix, by contrast, has about 1 in 14 at that level.

Apple appears to be running the HBO playbook: small slate, premium positioning, cultural impact over catalog scale. Which raises an interesting question given that HBO is currently for sale. Regardless of whether a Netflix or Paramount transaction ultimately succeeds, the new owner will face a strategic choice about what HBO becomes and Apple’s current approach offers one proof point for how a small-slate, high-budget strategy can work in the streaming era.

The L/XL Pipeline: A Lagging Indicator Turning Positive

Smaller productions recover first in any cycle - faster approvals, lower risk, quicker greenlights. L/XL projects take longer to develop and approve, so strength in that pipeline is a later-stage confidence signal.

What we’re seeing is encouraging: the incomplete H1 2026 dataset already shows 117% of the total L/XL projects that appeared in the final H1 2025 count. The slate is heavy with returning IP - The Bear Season 5, Severance Season 3, Daredevil: Born Again Season 3, Wednesday Season 3 - reflecting continued risk management through proven properties.

Context matters here. US domestic box office hasn’t returned to pre-pandemic levels of $11-12B; we’ve struggled to reach $9B in recent years, with significantly fewer wide releases. Studios are still solving for profitability in a world where theatrical economics have fundamentally shifted. The heavy reliance on sequels and returning IP in the L/XL pipeline isn’t lack of imagination, it’s rational capital allocation while the industry figures out what works.

So What Should Execs, Investors, and Operators Do With This?

For studio execs and producers: Geographic arbitrage is now table stakes, not an edge. If you’re not already optimizing for UK and Canada incentives, you’re late. The volume recovery is real, but the underlying economics remain challenged, expect continued scrutiny on mid-tier projects without clear distribution upside.

For investors: Production volume is a leading indicator for the service and infrastructure layer, post houses, VFX, equipment, sound stages. The Q1 signal suggests demand returning, with increased VFX demand expected in 2026 given the L/XL pipeline strength. Watch the pipeline conversion rate: if those 56 L/XL projects shoot on schedule, H2 2026 could look materially better. Treat the pipeline as a “conversion watch,” not a victory lap. The Netflix volume model generates steadier workflow for operators; Disney’s tentpole model is lumpier.

For operators in post, VFX, and infrastructure: Follow the geography. UK capacity is increasingly strategic; US-only positioning means serving a shrinking share of the market. A geographic shift isn’t just a production-office detail, it changes the economics and operational requirements for the entire ecosystem: connectivity, security posture, workflows that traverse borders, talent availability, tax-credit strategy, and vendor competition. The pipeline visibility is better than it’s been in two years, use it for capacity planning rather than waiting for bookings to arrive. And be realistic: the overall TV & Film production market remains compressed. Don’t staff for a 2022-shaped recovery.

The market is healing, but slowly, and the industry that emerges will be differently shaped, different geography, different studio mix, different risk appetite. The players who recognize the rebalancing and position for where production is going, rather than where it was, will be the ones who benefit as volume returns.

What I’ll be watching in the next SPI updates: does the Q1 uplift show up in the smoothed value (not just starts), and does the UK share keep climbing as the higher-budget projects convert? If both hold, 2026 may not feel like a “return to normal”, but it could be the start of a new normal that rewards the players who adapt early.

Welcome to the first edition of Storytellers & Silicon. I’ll be publishing twice monthly—once on production trends and market forces, once on broader industry perspective. If you’re in the business of putting magic on the screen, and navigating the technology that makes it possible, this is for you.

Every year, my team at Sohonet publishes a trends report. We aggregate data from our Screen Production Index, talk to customers across the studio and post/VFX ecosystem, and try to identify the forces that will shape how content gets made over the next several years.

This year’s report identifies six market forces that will define 2026-2030. You can read the full analysis on Sohonet.com. But I wanted to use this first newsletter to share what I’m actually thinking about—the things in this data that keep me up at night, or that I think the industry is underestimating.

The growth isn’t where you think it is

Large and extra-large productions are unlikely to grow in any meaningful sense going forward. The tentpole machine isn’t broken, but it’s not expanding either.

Where the six forces do have impact is in smaller and mid-sized productions—that’s where volume is increasing. And there are new sources of video production emerging as a direct consequence of those same forces: creator-economy professionalization, agency-backed micro-studios, AI-native content shops.

But here’s what makes this moment different from previous cycles: the workflow requirements of these productions are changing rapidly. It’s not just “more content”—it’s more content produced in new ways, with new tools, across more locations, by teams that look different than they did five years ago.

Disruption from AI and changing business models will likely define the next 18-24 months. If you’re planning infrastructure, talent strategy, or service offerings based on how things worked in 2022, you’re already behind.

Hybrid won, but nobody’s celebrating

The “cloud vs. on-prem” debate is over. Hybrid won. The HPA’s 2024 survey says 78% of productions now run hybrid workflows. Only 12% are all-cloud.

But here’s the thing: hybrid is harder to operate than either pure model. You’re managing on-prem storage, cloud burst capacity, and the connectivity between them—often across multiple continents as tax incentives scatter shoots further from traditional hubs.

The median project now touches 7-8 data nodes. Five years ago it was 3.

And cloud isn’t getting cheaper. AWS operating margins hit 31% in Q1 2025. EC2 GPU spot prices are up 17% year-over-year. The hyperscalers have no incentive to race to the bottom on pricing.

So hybrid is the flexibility play—you burst when you need to, optimize costs when iteration pressure drops. But it requires workflow architecture, automation, and integration that most of the industry is still building. The winners over the next five years will be the ones who figure out how to make hybrid operationally seamless, not just technically possible.

The creator economy is a parallel studio system

I’ve been in this industry for 30 years. I’ve watched a lot of “disruption” narratives come and go. But I think the creator-economy aggregation trend is real, and I think traditional Hollywood is underestimating it.

YouTube is now the dominant living-room platform for 18-40 year-olds. Not Netflix. Not Disney+. YouTube.

And the creator ecosystem is professionalizing fast. Publicis announced a 15-series influencer slate at Cannes. WPP opened a Creator Hub in London. Dentsu and IPG are launching equivalents. Goldman projects creator-economy revenue growing at 14% CAGR through 2028.

The pattern I’m watching: agency-backed micro-studios aggregating creators, moving production from bedrooms into professional hubs, and starting to need the same infrastructure that traditional productions use—real-time review, rapid iteration, integrated asset management.

This is a new customer segment for our industry. It’s also a new competitor for audience attention. The streamers know this. Netflix isn’t just competing with Disney+ and Max—they’re competing with MrBeast.

The question I can’t answer

Every force in our report points the same direction: more content, more locations, faster cycles, smaller teams doing more with AI assistance.

The part I can’t predict is how the talent equation resolves.

AI compresses costs by enabling smaller teams. But the work still requires skilled people—just different skills. Fewer junior rotoscope artists, more “AI wranglers” who can supervise and refine machine output. Fewer editors doing assembly cuts, more editors doing creative decision-making while AI handles the mechanical work.

Does that transition happen smoothly? Do the guilds adapt? Do training pipelines shift fast enough? Does the industry end up with a skills gap that slows adoption?

I don’t know. But it’s the variable I’m watching most closely over the next 24 months.

What’s next

Storytellers & Silicon will publish twice a month. The first post each month focuses on trends—what we’re seeing in the data, how the six forces are evolving, what’s changing in production economics. The second will be broader industry perspective—point of view on emerging tech, workflow shifts, business model changes.

If you’re in production, post, VFX, or platform operations, I hope this is useful. If you want the full 2026 trends analysis, it’s available at sohonet.com.

And if you have questions or see something in your corner of the industry that contradicts what I’m seeing—tell me. I learn more from being wrong than from being right.

— Chuck

Chuck Parker is CEO of Sohonet and serves on the board of the Hollywood Professional Association.

Subscribe now