Reka and Moonvalley Join Forces to Advance Models and Infrastructure for Physical AI

Reka, a prominent AI research laboratory dedicated to building foundational intelligence for real-world applications, has officially announced its strategic merger with Moonvalley. This significant collaboration is set to bolster Reka's capabilities by integrating Moonvalley's highly skilled team of AI researchers and engineers. The primary objective of this merger is to expedite the advancement of sophisticated AI models and the underlying infrastructure necessary for the emerging "physical AI era." The newly combined team will concentrate its efforts on developing AI systems capable of understanding, simulating, and interacting with the physical environment. Key to this endeavor is the World Language Action Model (WLAM), an innovative omni-model. WLAM is specifically designed to be trained on egocentric and other physical world data, enabling it to perceive and act effectively in real-world scenarios by performing realistic simulations for planning and execution. Mateusz Malinowski and Mikołaj Bińkowski, both former senior research scientists at Google DeepMind and key contributors to Google's Veo models, are joining Reka's research leadership as part of this merger. Their expertise in video generation and multimodal models is expected to be instrumental in Reka's mission. Dani Yogatama, CEO of Reka, emphasized that "Physical AI is fundamentally a research problem," highlighting the importance of this combined expertise in building AI systems that go beyond mere generation to truly comprehend and interact with the physical world. Malinowski further elaborated on their approach, stating that their models are built not just to generate video but to understand the mechanics of the physical world, including motion, physics, and temporal dynamics. This deeper understanding will allow robots to reason about potential consequences before acting, promoting safer and more responsible AI deployment. The unified strategy aims to integrate simulation and reasoning into a single, cohesive system, paving the way for advancements in autonomous defense, intelligent robotics, next-generation wearables, and media applications. The merger represents a concerted effort to tackle complex real-world challenges through advanced AI research and development.