The Man Who Walked Away from OpenAI and Raised $134 Million Instead
The way we build games is about to change more in the next five years than it has in the last thirty.
Not because of engines. Not because of platforms. But because of world models.
When I asked Pim De Witte, the founder of General Intuition, a company that recently raised $134M seed round, how games will be built in the future, his answer wasn’t another tired “AI will help with assets” take.
He described a creative revolution that moves from scripted to generative, and finally to agentic systems. In other words, games that think, act, and react.
In this world, the deterministic logic of Unity and Unreal won’t disappear, but it will be augmented by systems that behave less like machines and more like living organisms.
“A lot of what makes games great today,” Pim said, “is determinism. But LLMs and AI models are non-deterministic. They create variance, unexpectedness, and that’s what drives retention.”
Translated for the knuckledraggers like myself, this means, for example, that in a game, bots will be indistinguishable from players. Everything from friendly fire to flanking maneuvers, missed shots, or even panic retreats will feel human. This is because the AI won’t just follow rules, but rather understand the world it’s in.
The future of games might look less like code and more like jazz: structured enough to make sense, chaotic enough to keep us playing.
What Are World Models (and Why They Matter)
If language models predict words, world models predict what happens next.
In simple terms:
A video model predicts the next frame of a scene.
A world model predicts how the environment will react to your actions aka the next state of reality.
When you jump, shoot, or turn a corner, a world model doesn’t just guess what should be shown next; it understands why these actions are taken. It simulates causality, not just the visuals.
This is huge because every great game is built on cause and effect. You act, the world responds. From Pong to Minecraft, that’s been the essence of play. World models supercharge this by learning from billions of gameplay clips, understanding not just how worlds look, but how players behave within them.
The realization of the above is perhaps why Pim walked away from selling his previous startup, Medal.tv, even when OpenAI reportedly came calling with $500 million. He wanted to build AI born from games.
“Games are where humans go to express their intelligence at the highest level,” Pim said. “They’re the perfect merge of all modalities, text, geometry, perception, and action. That’s why we build intelligence through games, not language.”
Only an AI-Native Studio Can Reap The Full Benefits of AI. Right??
The short answer: no.
The real answer: you need to be AI-fluent.
Pim doesn’t believe legacy studios are doomed.
“If you have a solid engine, strong systems, and you start integrating AI where it helps, tooling, iteration speed, better agents, you’re already ahead,” he told me.
The myth that only “AI-native” startups can thrive is mostly branding. Yes, they move faster and prototype differently. But established studios hold something more valuable: taste, process, and culture. And those can’t be trained into a model.
In Pim’s words:
“AI is general-purpose. The best studios will use both the power of AI tooling and the depth of deterministic systems. It’s not about replacing what works. It’s about supercharging it.”
He’s right. Most of us aren’t losing the race to technology. We’re losing it to anxiety. The belief that someone else, somewhere, has the secret sauce. But there isn’t one. The winners will simply be the ones who start learning how to work with AI before it’s everywhere.
Why General Intuition Raised $133.7M
So why did investors pour $133.7 million into Pim’s new company?
Because General Intuition isn’t building AI to replace developers, it’s building AI to play with them.
Unlike OpenAI or Anthropic, GI isn’t obsessed with words; it’s obsessed with worlds.
Their models don’t predict text; they predict spatial-temporal outcomes: how things move, collide, and change. And thanks to Medal.TV’s library of over 2 billion gameplay clips they hold one of the richest datasets of human interaction ever recorded.
General Intuition is also a Public Benefit Corporation. A statement of intent that this technology should augment developers, not erase them. Pim’s mission is simple: keep gaming weird. Keep it human. Keep it the last corner of the internet where creativity thrives over algorithmic optimization.
A noble mission. But Pim must have missed the last decade in mobile gaming, where performance marketing triumphed over game design, and where the same old games held the crowns with ever more “creative” monetization systems.
Philosophical Musing: The AI Paradox We’re Living Through
If AI works as promised, society collapses with hundreds of millions unemployed. If it doesn’t deliver on the colossal hype and capital behind it, the markets implode under the weight of another bubble.
That’s not hyperbole. It’s the corner we’ve backed ourselves into. In just the last few months, UPS cut 48,000 white collar jobs, Amazon announced it will replace 600,000 jobs with robots, and white-collar layoffs have quietly spread far beyond Silicon Valley. At the same time, as much as 52% of college graduates are underemployed, with new grads entering a job market where “entry-level” roles increasingly mean “automated.”
The near-term outlook is harsh: junior talent is vanishing before it even begins, productivity is outpacing purpose, and capitalism’s efficiency drive is leaving little room for human inefficiency, also known as learning.
But Pim’s counterpoint is quietly hopeful:
“We’ve been outsourcing decision-making for decades, first to computers, then to networks. This is just the next phase.”
If history is any guide, AI won’t end human work; it will evolve it. The Industrial Revolution replaced the handloom with the factory, and in the long run, it created more opportunities than it destroyed. The same pattern might play out here, but only if we bridge the gap wisely.
That bridge will have to be built on adaptation, not resistance. The next few years will belong to the translators, the people who can take human creativity, empathy, and taste, and direct these new intelligent tools toward meaningful ends. We’ll have fewer manual creators, but more world builders. Fewer coders, but more orchestrators.
And gaming, ironically, might show us how. Games are safe laboratories for human progress. They let us test intelligence, coordination, and chaos without breaking the real world. Before we let machines run our economies, maybe we should first teach them to play nicely in a simulated one.
The short-term future may look like layoffs and anxiety. But the long-term prize, a world where intelligence is abundant and creativity is amplified, is still within reach. The challenge isn’t surviving the transition. It’s keeping our humanity intact long enough to design what comes next.
