When you accidentally knock a glass off a table, you know exactly what’s going to happen.
And you can immediately feel the consequences of your action, picture it in your mind and even predict where the glass might land.
That built-in sense of how the physical environment operates is something most people take for granted. Until recently, that type of reasoning was well beyond current generative AI systems.
But world models could change that.
And that shift will alter the marketing landscape, too.
What Are World Models?
World models are AI systems that grasp how the physical world works. They reason about the cause-and-effect logic of how things interact. Given an enormous range of scenarios, these systems can predict what will happen next.
They’re different from generative AI tools like ChatGPT or Claude, which use sophisticated large language models, or LLMs, to come up with the next best word or phrase. While LLM outputs can be surprisingly good, the models themselves have no real sense of what their predictions mean.
World models, however, are trained on the way objects move in space and time. As a result, they’re able to envision multiple possibilities for what an outcome might be.
Consider a virtual customer experience built using a world model. It could mirror the design of an actual store, for example, and then layer the environment with believable digital elements like an encounter with a real-looking synthetic influencer who might be selling the latest fashions from a retail brand.
And the customer could feel present and immersed in multiple realities at the same time.
The Metaverse Redux
While this scenario, with its immersive 3D spaces, virtual objects and digital layer of reality, may sound like the metaverse, there are significant differences.
Metaverse environments were built on scripted rules and rendered graphics. They had no persistent memory, no real physics, and nothing intelligent behind them. If you picked up a virtual object, it might float, glitch or slip through the floor. The technology wasn’t up to the promise.
World models can fix the elements the metaverse got wrong. They’re not rendering scripted scenes; they’re generating environments from a learned sense of how physical reality behaves in real time, including: cause, effect, gravity, object interaction, and so on.
That’s what makes the experiences coherent and persistent in ways the metaverse never delivered. Which means the creative canvas for marketers will get considerably wider.
And there’s no reason to believe world models would need cumbersome headsets that closed off the environment around you. Perhaps they’ll become the operating systems for self-driving cars, domestic robots or always-on augmented reality.
Meet the Players
The promise of what world models might bring has led to billion-dollar fundraising rounds for several startups. These include:
- AMI – Advanced Machine Intelligence was founded in late 2025 by Yann LeCun after he left his role of head of AI research at Meta. Rather than building on large language models, AMI is developing systems based on his Joint Embedding Predictive Architecture (JEPA). JEPA trains AI to predict how real-world environments will evolve by learning structured representations of physical reality based on video examples. AMI raised $1.03 billion in seed funding in March 2026.
- World Labs – Co-founded by Fei-Fei Li, a pioneer in AI research and computer vision, who’s widely credited with helping teach machines to see. Li coined the term ‘spatial intelligence’ to describe her company’s mission: building AI that can construct and navigate physical space, not just describe it. Their platform, Marble, generates spatially consistent 3D environments from a single image or text prompt — complete with physics, lighting, and persistent memory. Anyone can try it though results with a free account are limited. The company raised $1 billion to date.
- Google DeepMind has been active in this space since 2024 when it launched its Genie model. Genie 2, released in December 2024, can generate rich interactive 3D environments from a single reference image, complete with object physics, character animation, lighting effects, and even the behavior of other characters in the scene. It also has persistent memory, recalling parts of the environment no longer in view and rendering them accurately when they reappear. However, the system is currently in early testing and somewhat glitchy. Project Genie is currently available to Google AI Ultra subscribers in the US.
As you can see, world models are still in their early R&D phase. Yet, the companies focused on developing them share serious research depth and considerable expertise building AI. That’s part of why the field feels so promising.
What This Means for Marketers
Because world models aren’t ready for prime time, now is a good opportunity for marketers to begin imagining how these new systems could support their brands.
Here are a few directions we’re thinking about:
Immersive digital experiences.
Customers step into a space built around your product or service and interact with it in ways that weren’t possible before. World models know physics but exist in a digital space, so you could literally turn your environment on its head.
Surface to screen.
Everything from light fixtures, desks, buildings, parks, even the sky has the potential to becomes a real-time display. Imagine a flock of birds flying past you and transforming into a company logo.
Location-based storytelling.
If someone is standing next to a landmark while travelling, a holographic video offering a historically grounded story could pop up, brought to you by … a sponsoring brand.
Enhanced personalization.
Trying on clothes in a virtual environment powered by a world model means seeing the fit. And haptic sensors might simulate the feel which means fewer surprises when the package arrives.
How to Get Started
As your marketing team brainstorms out of this world concepts, science fiction offers a good source of future-facing ideas that can stretch your thinking before the technology fully matures.
So watch sci-fi movies and TV or read stories and books to get a sense of what’s possible … both positive opportunities and the more negative ones you’ll want to avoid.
This also helps you consider the privacy, data and ethical ramifications that could affect the reputation of your brand.
If you want to learn more about world models, check out this FMI podcast, made in partnership with our friends at Gordie, a startup that helps people turn their ideas into high-quality video clips and podcasts narrated by an AI interviewer.
This post was written by Martin Waxman, associate director, FMI.
Connect with FMI
Want to stay ahead of the marketing wave and prepare yourself for the future? Connect with us on our Website, LinkedIn, Instagram or Twitter/X.
If you enjoyed this newsletter, please share it and subscribe to receive it directly in your inbox.