Published 11 hours ago • loading... • Updated 9 hours ago
Reka and Moonvalley to advance models and infrastructure for physical AI
The merger adds former DeepMind, Meta and Google researchers as Reka builds models that can reason, simulate and act in the physical world.
On Thursday, June 11, 2026, San Francisco-based AI research lab Reka joined forces with Moonvalley, adding a team of researchers and engineers to accelerate development for the physical AI era.
This partnership aims to solve the research problem of "Physical AI," creating systems that understand and act in the physical world, according to Reka CEO Dani Yogatama.
Former Google DeepMind researchers Mateusz Malinowski, PhD, and MikoBaj BiDkowski, PhD, joined the lab, bringing expertise in video generation and multimodal models after contributing to Google's Veo.
The combined team will focus on the World Language Action Model , an omni model that simulates motion, physics, and temporal dynamics to enable robots to reason before acting.
Reka continues to operate four pillars, including Reka Labs and Reka Infer, powering complex real-world challenges from intelligent robots to autonomous defense and media applications.