Sooth Labs Raises Approximately $50M Seed at $335M Valuation to Build Long-Horizon Predictive AI Models

Prediction used to be a parlor trick. Now it is a balance sheet line item, and Sooth Labs just priced it like one. A roughly $50M seed round, led by Felicis Ventures, puts the Pittsburgh crew on the board at about a $335M valuation. Not bad for a company betting that the future is less about reacting fast and more about seeing around corners before the corner even knows it exists. Yann LeCun and Jeff Dean are in the mix as backers. Andrew Bosworth is advising. When names like that show up, it is usually not for the snacks.

The founding bench tells you exactly what game this is. Yaser Sheikh, Ruslan Salakhutdinov, Chuck Hoover, David LaRose, and Shih-En Wei come out of Meta AI and Carnegie Mellon with a bias toward hard problems and longer timelines. Yaser Sheikh, CEO, anchors the operation, and the tone is clear. This is not built for demos. This is built for decisions.

Sooth Labs is building long-horizon predictive models for geopolitical and market events. Not tomorrow’s stock tick. Not next week’s headline. The big stuff. The slow-moving, high-stakes, blink-and-you-miss-the-shift kind of events that decide who wins in finance, defense, insurance, and real estate. The arenas where being early is everything and being wrong is expensive.

The part people should not gloss over sits right beneath the surface. Most systems today chase speed and surface-level fluency. Sooth is leaning into time, probability, and interconnected systems. They are not just asking what happens next. They are asking what is likely to happen, why, and how confident you should be when you bet on it. That is a different muscle entirely.

And the lesson for founders watching from the sidelines is not just about pedigree, even though this team has plenty of it. It is about picking a problem that actually matters. High-stakes decisions, unclear futures, massive financial consequences. Then building something that speaks the language of those decisions. Not dashboards for the sake of dashboards, but signal in a world drowning in noise.

There is also a quiet strategy baked into this raise. Felicis Ventures does not just write checks, they place bets on inflection points. Pair that with operators and researchers who have already played at scale, and you get something that feels less like a startup guessing and more like a lab with a thesis.

Sooth Labs is not promising certainty. That would be a red flag. They are selling calibrated foresight. And in markets where a few percentage points of probability can swing billions, that starts to sound less like a feature and more like infrastructure waiting to happen.