Reka and Moonvalley Merge, Reshaping Physical AI Frontier
June 12, 2026, 9:39 am
Reka, a leading physical AI lab, has merged with Moonvalley. This move adds a team of top AI researchers and engineers. Their focus: accelerate foundational models for real-world intelligence. The combined effort targets the World Language Action Model (WLAM). WLAM will enable AI to reason, simulate, and act in the physical world. This propels advancements in robotics, autonomous defense, and next-gen tech. The merger strengthens Reka's mission to solve complex real-world challenges with advanced AI.
San Francisco's AI landscape just shifted. Reka, a prominent AI research lab, has joined forces with Moonvalley. This merger is strategic. It targets the burgeoning field of physical AI. The combined entity aims to build foundational intelligence. This intelligence will operate directly in the physical world.
The talent acquisition is significant. Moonvalley brings a robust team of AI researchers and engineers. Many previously worked at industry giants. DeepMind, Meta, Amazon, Microsoft, Google, Wayve, and Runway are all represented. This talent pool is deep. It provides Reka with critical expertise.
Two key figures stand out. Mateusz Malinowski, PhD, and Mikołaj Bińkowski, PhD, are now part of Reka's research leadership. Both were senior research scientists at Google DeepMind. They contributed heavily to models. These models later became Google’s Veo. Their arrival signals a major boost for Reka’s capabilities.
Physical AI is Reka's core mission. It involves developing AI that reasons. It simulates. It also acts in the physical world. This is not just theoretical. It addresses complex real-world challenges. Autonomous defense systems will benefit. Intelligent robots will evolve. Next-generation wearables will gain new capabilities. Media processing will see advancements.
The merger fuels an accelerated development timeline. Reka seeks to build advanced models. It also focuses on robust infrastructure. Both are crucial for the physical AI era. This era promises AI that truly understands its surroundings. It will interact with real-world dynamics.
A central project is the World Language Action Model (WLAM). This is an omni model. It stands as a unique approach. WLAM trains on specific data. This includes egocentric and other physical world datasets. The goal is clear: perceive and act in the real world. Realistic simulation is key for planning.
WLAM boasts advanced capabilities. It generates high-fidelity multimodal outputs. This means rich, diverse data representations. It also translates multimodal perception directly into robotic execution commands. Its spatial awareness is state-of-the-art. Its temporal awareness is equally advanced. This allows nuanced reasoning. It processes long-form video. It also handles streaming video in real-time.
Understanding how the physical world operates is paramount. WLAM moves beyond mere video generation. It grasps motion, physics, and temporal dynamics. This understanding is critical for robotics. Robots can then reason about consequences. They can plan actions safely. This approach fosters a more responsible AI future.
Unifying simulation and reasoning within a single system is the right path. This integration allows AI to predict. It enables AI to adapt. It grants AI a deeper understanding of its physical environment. This holistic approach is fundamental to Reka’s strategy.
Reka's operational framework supports this vision. The company functions across four distinct pillars. Each pillar contributes to the overall objective of physical AI dominance.
Reka Labs spearheads research and development. It builds foundational intelligence. This is the core engine for physical world interaction. Innovation is constant. Breakthroughs are sought.
Reka Infer provides the inference infrastructure. It offers an API for multimodal AI. This enables efficient deployment. It allows real-time processing of complex data. Speed and reliability are crucial here.
Reka Vision focuses on video processing infrastructure. It’s a platform designed for vast amounts of video data. It tags. It reasons. It searches. It clips. Managing large volumes of visual information is vital for physical AI.
Reka Claru handles training data. This pillar provides the necessary fuel for advanced AI models. It includes egocentric videos. Robotics trajectories are collected. World-model footage is curated. Expert human judgment is integrated at scale. High-quality data drives superior AI performance.
The future of AI is physical. It will move beyond screens and data centers. It will interact with our reality. Reka positions itself at the forefront of this revolution. The merger with Moonvalley amplifies this position. It brings together top minds and resources.
This new alliance promises significant advancements. Autonomous systems will become more reliable. Industrial robots will gain new intelligence. Everyday devices will offer richer, more intuitive interactions. The potential applications are vast. They span multiple sectors.
Reka continues to expand. The company actively seeks talent. Positions are open across research, engineering, and product development. This growth signifies commitment. It reflects ambition.
Enterprise partners show interest. They seek early access to Reka's platforms. These platforms offer novel capabilities. They promise competitive advantages. Direct engagement with Reka’s team facilitates this access.
Reka is headquartered in San Francisco. It was founded by researchers from DeepMind, Meta, and Google. This lineage speaks to its expertise. Its focus remains clear: building models and infrastructure for the physical AI era. This era is now unfolding. Reka and Moonvalley lead the charge. They build the future of intelligent systems.
San Francisco's AI landscape just shifted. Reka, a prominent AI research lab, has joined forces with Moonvalley. This merger is strategic. It targets the burgeoning field of physical AI. The combined entity aims to build foundational intelligence. This intelligence will operate directly in the physical world.
The talent acquisition is significant. Moonvalley brings a robust team of AI researchers and engineers. Many previously worked at industry giants. DeepMind, Meta, Amazon, Microsoft, Google, Wayve, and Runway are all represented. This talent pool is deep. It provides Reka with critical expertise.
Two key figures stand out. Mateusz Malinowski, PhD, and Mikołaj Bińkowski, PhD, are now part of Reka's research leadership. Both were senior research scientists at Google DeepMind. They contributed heavily to models. These models later became Google’s Veo. Their arrival signals a major boost for Reka’s capabilities.
Physical AI is Reka's core mission. It involves developing AI that reasons. It simulates. It also acts in the physical world. This is not just theoretical. It addresses complex real-world challenges. Autonomous defense systems will benefit. Intelligent robots will evolve. Next-generation wearables will gain new capabilities. Media processing will see advancements.
The merger fuels an accelerated development timeline. Reka seeks to build advanced models. It also focuses on robust infrastructure. Both are crucial for the physical AI era. This era promises AI that truly understands its surroundings. It will interact with real-world dynamics.
A central project is the World Language Action Model (WLAM). This is an omni model. It stands as a unique approach. WLAM trains on specific data. This includes egocentric and other physical world datasets. The goal is clear: perceive and act in the real world. Realistic simulation is key for planning.
WLAM boasts advanced capabilities. It generates high-fidelity multimodal outputs. This means rich, diverse data representations. It also translates multimodal perception directly into robotic execution commands. Its spatial awareness is state-of-the-art. Its temporal awareness is equally advanced. This allows nuanced reasoning. It processes long-form video. It also handles streaming video in real-time.
Understanding how the physical world operates is paramount. WLAM moves beyond mere video generation. It grasps motion, physics, and temporal dynamics. This understanding is critical for robotics. Robots can then reason about consequences. They can plan actions safely. This approach fosters a more responsible AI future.
Unifying simulation and reasoning within a single system is the right path. This integration allows AI to predict. It enables AI to adapt. It grants AI a deeper understanding of its physical environment. This holistic approach is fundamental to Reka’s strategy.
Reka's operational framework supports this vision. The company functions across four distinct pillars. Each pillar contributes to the overall objective of physical AI dominance.
Reka Labs spearheads research and development. It builds foundational intelligence. This is the core engine for physical world interaction. Innovation is constant. Breakthroughs are sought.
Reka Infer provides the inference infrastructure. It offers an API for multimodal AI. This enables efficient deployment. It allows real-time processing of complex data. Speed and reliability are crucial here.
Reka Vision focuses on video processing infrastructure. It’s a platform designed for vast amounts of video data. It tags. It reasons. It searches. It clips. Managing large volumes of visual information is vital for physical AI.
Reka Claru handles training data. This pillar provides the necessary fuel for advanced AI models. It includes egocentric videos. Robotics trajectories are collected. World-model footage is curated. Expert human judgment is integrated at scale. High-quality data drives superior AI performance.
The future of AI is physical. It will move beyond screens and data centers. It will interact with our reality. Reka positions itself at the forefront of this revolution. The merger with Moonvalley amplifies this position. It brings together top minds and resources.
This new alliance promises significant advancements. Autonomous systems will become more reliable. Industrial robots will gain new intelligence. Everyday devices will offer richer, more intuitive interactions. The potential applications are vast. They span multiple sectors.
Reka continues to expand. The company actively seeks talent. Positions are open across research, engineering, and product development. This growth signifies commitment. It reflects ambition.
Enterprise partners show interest. They seek early access to Reka's platforms. These platforms offer novel capabilities. They promise competitive advantages. Direct engagement with Reka’s team facilitates this access.
Reka is headquartered in San Francisco. It was founded by researchers from DeepMind, Meta, and Google. This lineage speaks to its expertise. Its focus remains clear: building models and infrastructure for the physical AI era. This era is now unfolding. Reka and Moonvalley lead the charge. They build the future of intelligent systems.