The Games Industry’s Three Body Problem
In physics, the three body problem describes a system where three celestial bodies exert gravitational force on each other in ways that make their movements essentially impossible to predict. Unlike the elegant simplicity of two bodies orbiting each other, add a third and the math breaks. Small changes cascade. Stable orbits become chaotic. The system lurches between configurations, sometimes for eons, before settling into a new equilibrium - or flying apart entirely.
Frank Rotman wrote a brilliant piece in 2022 applying this metaphor to venture capital. His argument: structural changes in the VC ecosystem (more capital, more startups, more LP demand) had destabilized the old equilibrium, and only firms that moved quickly to a new “stable point” would survive. Everyone else would spiral.
The games industry has its own three body problem. Three forces are pulling on studios simultaneously, and they are pulling in directions that are very hard to reconcile.
Force 1: The price ceiling
Gamers do not want to pay more for games.
This is not a vague sentiment. It is one of the most durable consumer preferences in entertainment. The standard price for a new game sat at $60 for roughly two decades. When publishers began testing $70 in 2020, it was met with grumbling. When Nintendo launched Mario Kart World at $80 in 2025, the reaction was outright hostility. Microsoft tried the same thing with The Outer Worlds 2 and walked it back within weeks.
Eight separate price increases hit consoles, accessories, and subscriptions in the first half of 2025 alone. And it’s not working. Gen Z spending on games dropped nearly 25% year-over-year, driven by inflation and a tightening job market. The average Steam user played just four new games in 2024 and 2025. Mid-priced titles like Clair Obscur and Split Fiction gained traction at least partly because they weren’t asking for $80.
Force 2: AI hostility
Players hate AI. Many developers increasingly hate AI too. And the numbers are stark.
A Quantic Foundry survey of nearly 1,800 gamers found that 85% hold negative attitudes toward the use of generative AI in games. 63% chose the most negative response option available. The researchers noted this level of rejection is rare in years of survey work — worse even than the reaction to blockchain gaming, which is saying something.
It’s not just players. The 2026 GDC State of the Industry report surveyed 2,300 game professionals and found that 52% now view generative AI as harmful to the industry, up from 18% just two years prior. Only 7% called it a positive force. Visual artists led the opposition at 64%, followed by designers and narrative writers at 63%.
At GDC 2026, Moritz Baier-Lentz of Lightspeed (whose portfolio includes stakes in both Anthropic and Epic Games) told a panel he was “shocked and sad” that the industry was “demonizing” what he called a “marvelous new technology.” The room did not agree. The same survey found that 28% of developers had lost their jobs in the past two years. When the people selling the technology express surprise that the people being displaced by it are unhappy, you have a communication problem at minimum and a political problem at maximum.
Studios that have used AI visibly have learned this the hard way. Even Larian faced backlash for AI-assisted concept art and Embark replaced AI-generated voice acting with human performers in Arc Raiders after launch. The lesson is clear: players and developers are watching, and AI is a reputational risk.
Force 3: Costs are still going up
None of this would matter much if games were getting cheaper to make. They are not.
Average AAA development budgets have increased roughly 8x since 2000. And the curve is accelerating, not flattening. Each generation of hardware brings diminishing returns in perceived visual quality but disproportionately higher costs to achieve them. Joost van Dreunen has been tracking this and makes a persuasive case that the trend is approaching a sustainability crisis.
Even outside AAA, costs are rising. Mobile games are cheaper to build relative to console/PC, but user acquisition costs have become so punishing that the marketing budget can dwarf development by an order of magnitude. Scopely spent over $1 billion marketing Monopoly Go — and still turned a profit, but that’s a deeply weird business model.
The trap
Here is what makes this a three body problem and not just a tough market.
The obvious solution to Force 3 (rising costs) is AI. Generative AI can meaningfully reduce production costs in asset creation, testing, localization, prototyping, and increasingly in code and design iteration. Studios that adopt it aggressively could, in theory, make games faster and cheaper.
But Force 2 says you can’t do this. Or rather, you can, but your audience and your workforce will punish you for it. An 85% disapproval rate is not a speed bump. It’s a wall.
And Force 1 says you can’t solve the problem from the revenue side either. You can’t make games more expensive to compensate for rising costs, because your customers won’t pay.
So: costs keep going up. Revenue per unit won’t go up. And the most obvious tool for reducing costs is politically radioactive.
That’s the instability. Every direction you push creates a reaction from the other two forces. The old equilibrium — make expensive games with large human teams, sell them for $60, rely on microtransactions for margin — is groaning under the weight.
Finding the Lagrange points
Rotman’s insight was that in an unstable system, a few positions of relative stability exist — what physicists call Lagrange points. These are positions where the gravitational forces balance out, allowing a small object to maintain its orbit without expending much energy.
I’d argue there are four stable positions emerging in games. Not all of them will work for every studio, and I could be wrong about some of them. But here’s my best guess.
Stable Point 1: Go small, stay human
Avoid the cost spiral entirely. Small teams, lower fidelity, no AI controversy, price at $20-30 where consumer resistance evaporates. Two of the top 10 most played games on Steam at any given time are low-fidelity games made by small teams (Everwind and Slay the Spire 2 at time of writing). This is not a consolation prize. This is where some of the best returns in the industry are coming from.
I’ve written about this before as minimum viable fidelity. The idea is simple: figure out the lowest level of visual polish you can get away with to attract your target audience, and invest your savings in gameplay depth and iteration speed.
This position is stable because it renders all three forces irrelevant. You’re not spending $200M, so you don’t need AI to cut costs. You’re not charging $80, so price resistance doesn’t apply. And you’re not using AI in your production, so nobody’s angry at you.
The catch: you need a genuinely great game to break through at this level, and even that is not enough. There is no marketing budget to buy attention. Your game has to be its own distribution. In 2025, over 19,000 games launched on Steam. The median sold 500 copies.
Stable Point 2: AI in the pipes, not on the screen
Use AI aggressively in the production pipeline — code generation, testing, asset iteration, localization, internal tools — but keep it invisible to consumers. The studio equivalent of using AI for back-office work while keeping the customer-facing layer human.
This is probably where most large studios will end up, even if they won’t say so publicly. 36% of game companies surveyed already use generative AI tools in some capacity, but usage skews heavily toward management and business professionals rather than the artists and designers whose work players actually see.
The political calculus is straightforward: players don’t care how you QA tested your game or generated your internal documentation. They care whether the art looks like it was made by a person who gave a damn. Use AI where it’s invisible, and invest the savings in the parts players do see.
The risk is that this is an unstable secret. The moment a leak reveals the extent of AI use behind the scenes, you face the same backlash as studios that used it visibly. Ask any studio that’s been caught.
Stable Point 3: AI-native, with permission
Build for the subset of gamers who actively want AI-driven experiences. Procedural worlds, personalized narratives, emergent systems, AI dungeon masters. This is Rotman’s “non-consensus alpha” equivalent — you’re betting against the 85% disapproval number and finding the 15% who are curious or excited.
There’s a real audience here. The Quantic Foundry data showed that gamers motivated by power progression (leveling up, gear upgrades) were the most positive toward AI. And a vocal minority of players — especially older gamers — are genuinely excited about AI-enabled experiences like infinite story branching or deeply personalized game worlds.
The bet is that 85% hostility today doesn’t mean 85% hostility in three years. Consumer sentiment toward new technologies tends to follow a pattern: initial rejection, gradual normalization as quality improves, eventual acceptance. The same gamers who hated free-to-play in 2008 now spend freely on battle passes.
This may not happen. The anti-AI sentiment in gaming feels is deeper rooted than, say the resistance to F2P - it’s wrapped up in concerns about job displacement, artistic integrity, and the broader cultural backlash against AI that’s happening across many industries. Maybe it softens. Maybe it calcifies. I genuinely don’t know.
Stable Point 4: Platform and UGC
Offload content creation costs to players. Roblox, Fortnite Creative, Minecraft, and increasingly any game with robust modding tools.
In this model, AI becomes a creator tool rather than a replacement for creators, and that changes the politics entirely. A player using AI to build their dream Roblox experience is not threatening anyone’s job. They’re extending the life of a platform. The studio’s cost structure shifts from “make all the content” to “build the tools and the marketplace,” which is a much more scalable and defensible business.
This is the position that best resolves all three forces simultaneously. Development costs are shared across a creator ecosystem. Prices are flexible (free-to-play platform, paid creator content, take rates). And AI is welcomed because it empowers players rather than replacing developers.
The constraint: this only works for certain kinds of games. You can’t build a UGC platform out of a narrative RPG. And the platform positions are increasingly locked up by incumbents with massive network effects.
Unstable orbits
If you’re running a game studio and you can’t see yourself clearly in one of these positions, Rotman’s framework suggests you’re in trouble. The death spiral for games studios looks a lot like the one he describes for VC firms: miss targets, struggle to raise capital (or generate revenue), lose talent, ship worse products, miss targets again.
The studios most at risk are mid-sized teams making mid-budget games that need to charge full price and can’t easily reduce their cost base without AI but can’t use AI without blowback. That’s a lot of studios.
What I don’t know
I want to be honest about the limits of this framework. The three body problem is a useful metaphor, but metaphors simplify. The real system has more than three forces — platform economics, regulatory risk, cultural shifts in what “gaming” means — and the stable points I’ve described might turn out to be less stable than I think.
The biggest uncertainty is Force 2. Will gamer hostility toward AI soften as the technology improves, the way hostility toward free-to-play and DLC softened? Or is this different — more durable, more values-driven, more permanent? If it softens meaningfully, Stable Point 3 becomes much larger and Stable Point 2 becomes much easier to maintain. If it doesn’t, the cost pressure on studios without a clear path to profitability gets worse every year.
I don’t have an answer. But I think the framework is useful for at least asking the right questions.