Inside the shift toward data-native markets

Matt Ober is a managing partner at Social Leverage. Matt was most recently the chief data scientist at Third Point, where he built the data analytics and technology platform used to enhance the firm’s investment capabilities in equity, structured credit, venture capital and cryptocurrency.

One of the biggest takeaways from 2025 is that markets are becoming data-native. Consumer culture, trading infrastructure and AI development are converging around the same principle: the richer the data, the more valuable the market built on top of it. Sneakers start acting like financial assets once pricing data matures. Brokerages pull prediction infrastructure in-house once user flows become predictable at scale. World models demand entirely new forms of sensory data to function. In short, finance, culture and AI are beginning to operate on the same underlying architecture.

Here's how this is happening in real time.

1. Vertical integration and the new era of financial infrastructure

The first signal appears inside financial infrastructure itself.

Robinhood continues to double down on winning products. Their prediction-markets announcement is their verticalization play to own the underlying infrastructure. Analysts quickly recognized the significance of the move. Many noted that Robinhood’s work with Susquehanna International Group to acquire a clearinghouse and stand up a fully integrated prediction market marks a meaningful step toward controlling more of the pipes themselves. Others pointed out that Robinhood and Susquehanna are now building a CFTC-licensed exchange and clearinghouse, a shift that helps clarify the company’s long-term direction and competitive ambitions.

What does this mean for their Kalshi partnership? When they announced $100 million in revenue in under 12 months from prediction markets, and noted that more than half of Kalshi’s flow came from Robinhood, it was already a strong sign that Robinhood would eventually bring the product in house. Do they keep the Kalshi partnership? Is there a Polymarket partnership potential? Time will tell.

Robinhood continues to make smart acquisitions. It seems like they are winning in financial services and nobody is even close. They acquired a credit card startup. They acquired pieces for their wealth management strategy including TradePMR and they have made many other savvy moves.

They are taking the type of actions that are getting investors excited for the long term and keeping everyone on their toes for what is next.

Pulling infrastructure in-house is becoming a pattern. Prediction markets generate valuable data that compounds with scale, and the companies closest to that flow are now moving to internalize it.

2. Consumer culture is turning into financial market structure

A second shift shows up in culture, where data density is turning hobbies into investable markets. In 2021, Bloomberg published a piece that framed sneakers not as fashion but as an asset class.

Nike at the time was booming. Every sneaker on their SNKRS (Sneakers) app sold out, and then StockX, GOAT (Greatest of All Time), eBay, and dozens of other second-hand sites were selling the sneakers for anywhere from two to twenty times the price. Then the bust happened. Sneakers, collectibles, Pokémon, trading cards and everything in between came down to earth. Maybe it was Covid, maybe it was stimulus, maybe it was low interest rates, but whatever the cause, it ended.

Now Kalshi, the prediction markets platform, has partnered with StockX to bring that world back, with predictions on everything from sneaker prices to Labubu dolls to other collectibles. It seems likely that Polymarket will follow. Does Robinhood add this to their app alongside sports and economic prediction contracts?

Neustreet and Pricing Culture were two early data companies in the collectibles space, and you would think the value of their data just skyrocketed with this new use case and demand. Sometimes for startups and companies, the priority is simple: survival. For data companies it has always been true that the value of their data goes up once you have years of history. Now that prediction markets are coming for collectibles, it will be interesting to see who the winners are. StockX is great, but it is one platform and does not give you a true view of the world or the price of collectibles. You would think a better partner is someone aggregating across marketplaces.

One thing is clear, prediction markets need data. They need to keep innovating, and they want to find more users. Collectors and sneakerheads are a great way to bring in more users and bring the excitement of collecting back.

Once pricing data becomes reliable and liquid, the distinction between a consumer good and a financial asset narrows. Prediction markets are accelerating that crossover.

3. The next frontier: world models and the race for real-world data

The deepest transformation comes from AI, where models now require data that reflects the physical world itself.

World models learn by watching video or digesting simulation data and other spatial inputs, building internal representations of objects, scenes and physical dynamics. Instead of predicting the next word, as a language model does, they predict what will happen next in the world, modeling how things move, collide, fall, interact and persist over time. The goal is to create models that understand gravity, occlusion, object permanence and cause-and-effect without being explicitly programmed to do so.

It makes me think that humanoid robots, other robotics platforms, and autonomous vehicles will be some of the biggest beneficiaries of world models.

So how does this relate to data?

Data is one of the key challenges. Those building large language models have been able to get most of what they need by scraping the breadth of the internet.

World models also need a massive amount of information, but from data that is not consolidated or as readily available.

One of the biggest hurdles in developing world models is their need for massive amounts of high-quality multimodal data that captures how agents perceive and interact with physical environments, Encord president and cofounder Ulrik Stig Hansen said.

Encord offers one of the largest open source datasets for world models, with 1 billion data pairs spanning images, video, text, audio and 3D point clouds, along with a million human annotations collected over months.

But even that is only a starting point. Hansen said production systems will likely need far more.

AI is moving fast. There is not a day that goes by without some sort of news coming out. The underlying fact is that data is one of the key pillars of all of this AI innovation. It now seems like the world model data needs will be the next area of attention.

As AI changes from predicting language to predicting physical reality, structured multimodal data becomes the new scarce resource. Companies that can capture or generate it gain an advantage that is hard to replicate.

The bigger picture

Across finance, culture and AI, the same pattern is showing up. As data gets better, markets shift. Companies pull more of the stack inside once they see how their data really works. And as AI starts to depend on real-world information, what counts as valuable data keeps expanding. The outcome is clear: markets are being rebuilt from the ground up around data.