Introducing MolmoAct: A Revolutionary Embodied AI Model for 3D Reasoning

Ai2 has unveiled MolmoAct 7B, a new class of AI model that reasons in 3D space and brings intelligence to the physical world. The model uses visual reasoning tokens to transform 2D image inputs into 3D spatial plans, enabling robots to navigate the physical world with greater intelligence and control. MolmoAct is trained entirely on open data and prioritizes transparency and openness, making it easy to preview and steer robot behavior in real time. It is the first in a new category of AI model called an Action Reasoning Model (ARM), which interprets high-level natural language instructions and breaks them down into a transparent chain of spatially grounded decisions.

NVIDIA (NASDAQ: NVDA) has made significant strides in the robotics sector by unveiling a suite of new technologies designed to revolutionize the development and deployment of robotic systems. The company's latest innovations include the Omniverse NuRec libraries, Cosmos physical AI models, and enhanced AI computing infrastructure. These advancements aim to address key challenges in robotics development, such as accurate world reconstruction and physical AI reasoning.

The Omniverse NuRec libraries feature 3D Gaussian splatting, a technique that allows for more realistic world reconstruction. This capability is crucial for developing robots that can navigate real environments effectively. The introduction of Cosmos Reason, a 7-billion-parameter reasoning model, is particularly noteworthy. Unlike general-purpose vision language models (VLMs), Cosmos Reason is specifically designed for physical AI, enabling robots to reason about the physical world using common sense and physics understanding. This capability is essential for robots to function reliably in unstructured environments.

NVIDIA has also announced the launch of RTX PRO Blackwell Servers and expanded DGX Cloud availability on Microsoft Azure Marketplace. These new hardware and cloud solutions are designed to support the most demanding workloads in robotics and physical AI development. The company has partnered with major industry players such as Amazon Devices & Services, Boston Dynamics, and Figure AI to integrate these technologies into their respective platforms.

The new solutions aim to transform robotics development by combining AI reasoning with physically accurate simulation. This approach positions NVIDIA as the infrastructure provider for the entire robotics development pipeline. The company is fostering community adoption by making Isaac Sim 5.0 and Isaac Lab 2.2 available on GitHub, and the OpenUSD Curriculum initiative further strengthens NVIDIA's ecosystem leadership position in 3D simulation.

The 2 million downloads of Cosmos WFMs indicate strong developer interest in these new technologies. By making these tools and models accessible, NVIDIA is creating a complete ecosystem where developers can build, train, and deploy robotic systems. The partnerships with industry leaders validate market demand and create potential revenue streams across multiple verticals.

NVIDIA's announcement represents a significant expansion of its robotics technology stack with three key pillars: improved simulation capabilities, advanced AI models, and purpose-built computing infrastructure. These innovations position NVIDIA to capture significant value as the robotics industry grows and matures across manufacturing, autonomous vehicles, and service robots.

References:
[1] https://www.stocktitan.net/news/NVDA/nvidia-opens-portals-to-world-of-robotics-with-new-omniverse-m3l0104hz8jn.html