(Memesplode)

"why do videogame budgets keep going up even if no one cares about next gen graphics anymore"

So the premise is wrong but in an interesting way. The premise is that the budget escalation is being driven by "next gen graphics" — meaning, presumably, the rendering pipeline, the lighting model, the polygon ceiling, the resolution target, all the stuff a hardware reviewer talks about when a new console comes out. And the correct response to that premise, the one I see in every YouTube comments section, is that diminishing returns kicked in around the PS3 era and now we're spending fortunes for marginal gains nobody asked for.

That argument is true at the level of the rendering pipeline. It's almost completely irrelevant to the budget question.

Because — and this is the thing the discourse never quite grasps — the cost of a modern AAA game isn't being driven by the cost of one high-fidelity asset. It's being driven by the cost of all the high-fidelity assets, and by the implicit contract with the player that everything you can see has been hand-finished to the same standard as the hero stuff. Once the fidelity ceiling rose, the floor rose with it. And the floor is where almost all of the labor lives.

A 2002 game could have a hero asset (the protagonist's gun, say) that was lavishly textured and animated, and surround it with environment props that were essentially decorated boxes — a barrel was a cylinder with a barrel-texture wrapped on it, and that was fine, because the rendering pipeline couldn't have shown you a real barrel anyway. The fidelity of the hero asset and the fidelity of the trash asset were both being filtered through the same low-resolution funhouse mirror, and the player's eye couldn't catch the discrepancy.

Modern photoreal rendering doesn't have that mercy. If your hero asset is a hand-modeled, hand-textured, PBR-shaded, ray-traced, motion-captured character, and your barrel in the corner of a Vice City alley is a low-poly cylinder with a flat texture, every player will see the seam. So you can't ship the cylinder. You have to ship a hand-modeled barrel with realistic wood grain, weathering, label decals, displacement maps, ambient occlusion, the works. Times every barrel in the game. Times every prop in the game. Times every NPC in the game.

This is the part that keeps surprising people who think the cost is in the lighting model. The lighting model was a one-time engine investment. The content is the bottomless pit.

Game Developer's Raph Koster did some back-of-envelope math a few years ago — comparing AAA productions across decades, normalizing for inflation — and the punchline was that cost-per-megabyte of content has stayed weirdly flat over thirty years, because "salaries are salaries." It's just that we now expect games to ship with vastly more megabytes than we used to, and each of those megabytes was made by a person.

Red Dead Redemption 2 has, depending on how you count, somewhere in the neighborhood of 200,000 separate assets and roughly 500,000 lines of recorded dialogue. That's not because Rockstar is gilding the lily on its lighting solver. That's because the implicit contract is that you can wander into any building in Saint Denis and the building will have an interior, and the interior will have furniture, and the furniture will have books, and a book will have a title, and the title will be readable.

That contract scales by a multiplier-of-multipliers. Somewhere on the floor of Rockstar North between 2019 and 2026, a thousand artists were modeling individual newspaper layouts.

(There's a weirdly poignant industry term for this: "asset density." The phrase is doing a lot of moral work — it's how you say "we expect the artist to fill every square foot of this map with handmade objects" without saying anything that would attract organized labor.)

So that's the technical answer, which is mostly a content-volume answer dressed up as a graphics answer. But it's not the whole answer, because the technical answer doesn't explain why studios with completely different fidelity targets are also burning through money at unprecedented rates.

Concord wasn't trying to push the rendering envelope — it was a stylized hero shooter with cartoon proportions and flat lighting, and it cost somewhere between $150 million and $400 million depending on whose accounting you trust, for an eight-year development cycle that culminated in a two-week commercial run before Sony refunded everyone and shut the servers down.

You cannot blame the rendering pipeline for that. The rendering pipeline barely costs anything. Concord burned through nine figures because of three other things, and those three things explain most of the AAA budget escalation more honestly than any "next-gen graphics" argument can.

First, team size. A 2001 AAA game shipped with a credits roll of maybe sixty people. A 2025 AAA game has credits in the thousands, often spread across twelve or fourteen studios on three continents.

This is partly because content volume scales linearly with people but mostly because the category of jobs has multiplied. There's now an entire department for live-service infrastructure (anti-cheat, telemetry, moderation, server reliability engineering) that didn't exist in 2001 because the game was a CD-ROM, not a service. There's a department for monetization design. There's a department for community management. There's a department for accessibility compliance. There's a department for localization-as-cultural-adaptation rather than localization-as-translation. None of these existed and all of them exist now and all of them have headcount.

Second — and this is the part of the budget that no marketer ever talks about and every accountant lives in fear of — the operational tail.

A modern AAA game isn't a product, it's an annuity, and the annuity has running costs forever. GTA Online has been live since 2013 and has received nearly continuous content updates for thirteen straight years, which is roughly six full development cycles by 2001 standards stacked end to end. When you see "GTA VI cost three billion dollars," part of what you're seeing is the embedded cost of building the live-service engine that will support the next thirteen years of content drops, which will themselves cost an additional several hundred million per year to operate. The published budget number is the iceberg's tip; the iceberg keeps growing after launch.

Third, marketing. This is the one most people are dimly aware of but underestimate dramatically.

The standard rule of thumb in the industry — repeated by Bobby Kotick in earnings calls for two decades and never seriously contradicted — is that marketing spend on a tentpole AAA release should roughly equal development spend, sometimes exceed it. Modern Warfare 2 in 2009 spent $50 million on development and $200 million on marketing. The $200 million wasn't a vanity number. It was the calculated cost of making sure that on launch week, every man, woman, and gaming-adjacent child in North America had been impression-bombed with the title enough times that not buying it felt like declining a social invitation.

The marketing isn't selling you a game. It's selling you the obligation to participate in a cultural event, which is structurally different and significantly more expensive.

(Aside: this is also why the price of a marquee release has barely budged in twenty years even as costs have multiplied. The publisher cannot afford to price out the marginal buyer, because the marginal buyer is the entire point of the marketing investment. A $90 GTA VI is already getting pushback. A $120 GTA VI would mathematically be more rational and would also collapse the cultural-event logic that the whole apparatus depends on.)

Now. Having said all that, here's the part of the answer that almost nobody in the industry will tell you out loud, even though it explains more of what's actually happening than any of the technical or operational stuff combined.

The budget keeps going up because nobody at any executive level inside a major publisher is incentivized to be the person who said "let's spend less than last time."

The career math is brutally asymmetric. If you green-light a $300 million project that earns $700 million, you are slightly above replacement-level and might keep your job. If you green-light a $150 million project that earns $700 million, you are also slightly above replacement-level — but with the added asterisk of "this guy doesn't believe in the company's products enough to spend on them," and you are first in line for layoffs the next time the stock dips.

The asymmetry runs the same direction for marketing budgets, for headcount, for studio acquisitions. The only career-safe move is to spend more than the previous guy. If the project succeeds you can claim the spend was necessary, and if the project fails you can claim you spent everything that could reasonably have been spent and the failure was due to "market conditions."

This is the Bobby Kotick equilibrium and it is shockingly stable. It will keep being stable as long as the industry's tentpole hits keep returning enough on capital to mask the failures.

GTA VI is going to make the math work for everybody for at least another five years, and during those five years, every other publisher will be staring at GTA VI's budget and quietly raising their own to match, because the only readable signal in their environment is "the most successful franchise in history is spending three billion dollars per release," and the only career-safe response is to spend something that approximates that.

(The other thing this dynamic explains is the eight-year cycle for Concord. A different company, with different incentives, would have killed the project at the 2023 alpha review when a source described it as being in "horrible shape." Sony didn't kill it. Sony spent another $200 million on it.)

(Why? Because the executive who had championed Concord internally as "the future of PlayStation" had built so much career capital on the project that killing it would have meant publicly killing himself, and the institutional reflex in any large organization is to shovel money at a sunk cost rather than admit that the cost was sunk.)

The $200 million in additional spend wasn't a calculation. It was a face-saving operation, and face was worth $200 million because the alternative was Herman Hulst standing in front of a board and saying "I bought a studio for hundreds of millions of dollars and the studio's flagship product is unreleasable."

This kind of thing is not unique to video games. It just looks more visible there because the project lifecycles are short enough to read in real time.

There's a useful comparison case here from a completely different industry, which is the late-1970s American film business. After the runaway success of Jaws and Star Wars, Hollywood studios convinced themselves that the way to print money was to greenlight increasingly enormous productions with marquee directors and unlimited shooting schedules — Heaven's Gate, Apocalypse Now, 1941, the productions that took five years to make and went five times over budget on the theory that the next blockbuster was always around the corner.

Most of those productions failed. The ones that succeeded — Apocalypse Now, eventually, mostly — did so on terms that bankrupted the studios anyway.

United Artists collapsed. The model collapsed. The industry restructured around tentpole franchises with rigorous cost discipline, and that's roughly the regime we still live under in film fifty years later.

The video game industry is somewhere around 1979 in this comparison. The Bobby Kotick equilibrium has not yet hit its Heaven's Gate. Concord might be a leading indicator. So might the wave of cancellations of mid-budget projects that have been quietly killed at Microsoft and EA in the last two years, casualties of the same logic that kept Concord alive past its expiration date — once a project crosses some threshold of internal political investment, killing it is harder than continuing it, even when continuing it costs hundreds of millions more.

(The entire concept of a "mid-budget AAA" has basically disappeared from the industry, and not because of consumer preference. The consumer is fine with mid-budget games — see the entire indie sector. The mid-budget AAA disappeared because no executive can afford the career risk of greenlighting one. A $50 million project either earns enough to be a yawn or fails and is held against you forever, while a $300 million project that fails can be blamed on "market conditions." The variance asymmetry pushes everyone toward the high end.)

So when people ask why budgets keep going up, they're usually asking a technical question — why aren't the diminishing returns of photorealism causing the curve to flatten? — and getting a technical answer that doesn't satisfy them, because the curve isn't being driven by photorealism in the first place.

It's being driven by content volume scaling at modern fidelity, by team-size growth across functions that didn't exist twenty years ago, by the live-service operational tail, by marketing-as-cultural-event, and by the brutal career incentives that make underspending professionally lethal at every level of every publisher.

The graphics are doing a small part of the work and getting all the blame, the way the visible part of any system always gets all the blame.

The Atari 2600 era was different in this respect, because in the Atari 2600 era a game was made by one guy in a closet and the budget was his salary for six months and the cultural-event apparatus didn't exist — the cultural event in 1981 was the category of video games, not any specific title.

By 2001 the apparatus existed but was still mostly aimed at game magazines and TV ads during cartoons. By 2010 the apparatus was global, integrated, and dependent on the kind of cross-platform marketing budgets that only a handful of publishers could afford. By 2020 the apparatus had eaten the industry, and you could no longer make a tentpole AAA release without participating in it, and participating in it cost as much as making the game itself.

This is also, incidentally, why the indie sector has been having such an extraordinary creative decade. The indie sector has opted out of the cultural-event apparatus entirely.

Hades cost a few million dollars and shipped on Steam and made its money back in three months on word of mouth. Balatro was made by one guy. Vampire Survivors was made by another guy. These games are not in competition with GTA VI for any resource other than the player's time, and the player's time is plentiful and getting more plentiful as the AAA tentpole release calendar gets thinner and slower under the weight of its own budgets.

There's even a case to be made that the AAA budget escalation is creating the indie boom, by starving the mid-tier and pushing both creative talent and player attention toward the only segment where actual experimentation is still allowed. Every Rockstar North programmer who quits to make a cozy farming game with three friends is a small protest vote against the apparatus, and the apparatus has been generating more of those protest votes every year.

Same as it ever was.

chemists: don't think they know enough to be an authority in other fields

biologists: aren't even sure they know enough to be an authority in their own firld

now, your computer inside the bunker has a game-inside-a-game on it which is a charming farming sim of undeniably greater quality and scope than the survival game you're playing. therefore, the object of the game becomes to keep your bunker secure so you can play the farming game more.

now, once you achieve the highest rating in the farming game, a secret shop inside it unlocks, and one of the novelty items you can purchase is a game console, giving you access to games-inside-a-game-inside-a-game. most of the games for it are typical mobile shovelware, but one of them is a highly polished, extremely brutal precision platformer with amazing level design and production values exceeding that of the survival game and farming sim combined.

Meanwhile Faye:

Source

"A marriage ending isn't a failure at all. I spent eleven years with her. We were so in love that we couldn't image life apart from each other. We got our own place, adopted a dog, and supported each other through school. I thought if tow people loved each other enough the rest would fall into place, except... love isn't everything.

And I didn't want to believe that, but we were sitting in counseling one day, talking about our future and I realized we were describing two completely different lives. Where we'd live, what kind of life we wanted, what made us happy. And it hit me that- I love this woman and this woman loved me. And after eleven years of loss, grief, career changes, we were so deeply in love... but we weren't aligned. And I kept thinking 'We just need to try harder. We can find some compromise to make this work,' because that's what you're supposed to do when you love someone, right?

But the reality was, we had just become different people. Her trade school took her in one direction, my graduate degree in another and trying to force us back into who we were five years ago wasn't coming from a place of love. It was coming from a place of fear. Fear that, if this ended, it meant we wasted eleven years. But sitting there across from her, I realized: That's not how love works.

Those eleven years happened. They were real. The dog, our home, showing up for each other through grad school and trade school. I wouldn't change a single thing because loving someone doesn't mean you're meant to stay with them forever. And letting go doesn't erase what you had. We measure marriage by whether it lasts forever or not, but what if we measured it by whether it mattered?

What if we measured it by the love we gave, the life we built, and the people we became? Because love's job isn't to last forever, it's to help you become fully completely yourself, and sometimes the most loving thing you can do is give each other permission to be yourselves, separately. But the dog doesn't know were' divorced. He just gets two Christmases now."