#USA #San Jose #Robotics #Nvidia GTC #Physicl

Physicl Emerges at NVIDIA GTC with a Game-Changing Infrastructure

On March 17, 2026, during the NVIDIA GTC in San Jose, Physicl unveiled an innovative data infrastructure platform specifically targeted at Physical AI and robotics. This launch marks the company’s first public appearance, transitioning from stealth mode to bringing essential technologies to the forefront of AI development. Founded by a team with extensive experience, including previous projects like Nfinite, Physicl aims to enhance the interaction capabilities of AI across spatial dimensions, a timely endeavor as advancements in artificial intelligence pivot from traditional language and image processing systems to more sophisticated, interactive models.

Bridging the Data Gap for Advanced AI Models

Physicl recognizes a critical bottleneck in the evolving AI landscape— the need for structured, physics-aware, simulation-ready 3D data. Alex de Vigan, the CEO of Physicl, states, “Every major advancement in AI has demanded a new data layer,” emphasizing the necessity of having data that accurately reflects real-world physics and spatial relationships. The comprehensive foundation laid by Physicl will empower robots and AI models to understand their environments, thereby enabling reliable operation in real-world situations.

Multi-Faceted Pathways Driving Physical Intelligence

Physicl's platform caters to several converging fields of physical intelligence:

• - Robotics: By providing physics-accurate simulations for training embodied systems, Physicl allows these entities to adeptly navigate and manipulate their environments.
• - World Models: The platform provides structured 3D environments essential for developing spatial reasoning and causal understanding.
• - Vision-Language Models (VLMs): By grounding multimodal systems in simulation-ready data, Physicl supports the next generation of AI capabilities that harmonize visual inputs with linguistic contexts.

With nearly ten years spent digitizing the physical world, Physicl’s data infrastructure ensures steady, production-grade data pipelines, moving away from sporadic datasets and towards scalable systems ready for global readiness.

Collaboration with Industry Leaders

To further enhance its platform, Physicl has joined forces with prominent technology firms such as Meta, DeepMind, and Getty Images. These collaborations bolster the three significant infrastructure components offered by Physicl:

• - Data Normalization: This feature converts various visual inputs into consistent, spatially structured 3D representations.
• - Physics-Aware Data Augmentation: It allows the generation of extensive, domain-randomized datasets grounded in realism, ensuring the data remains IP-safe.
• - DataSim Pipelines: This function facilitates the creation of environments suitable for training embodied AI and sim-to-real transfer processes.

Leveraging NVIDIA's Advanced Technology

Physicl optimally utilizes NVIDIA's technology stack designed for Physical AI, including tools like NVIDIA Omniverse and Isaac Sim. These tools are pivotal for developers needing high-quality 3D environments for their model training and evaluation processes, a task that is often time-consuming and expensive.

The infrastructure offers:

• - Omniverse-ready Assets: Physically realistic, simulation-ready 3D environments that seamlessly integrate with existing workflows.
• - Isaac Sim and Lab: Tailored environments that support robotic training for manipulation, navigation, and complex tasks.
• - Cosmos-Compatible Data: 3D structures that align with foundational world model requirements.

Launching with a rich catalog of millions of simulation-ready assets, Physicl is poised to provide continuous access to essential data, akin to how Getty Images offers licensed visuals.

Setting the Stage for Future AI Developments

Physicl’s innovative approach reflects a growing consensus in the tech community: the next generation of AI demands not just larger volumes of data but an entirely new paradigm that acknowledges the interplay of geometry, physics, and spatial relationships. “NVIDIA has established the infrastructure needed for Physical AI,” notes de Vigan. “Physicl is dedicated to being that vital data layer propelling the forthcoming wave of AI and robotics innovations.”

During the NVIDIA GTC, attendees can explore the Physicl platform at Booth 3307, witnessing firsthand how simulation-ready assets can enhance AI and robotics development. For developers and researchers interested in early beta access to this groundbreaking platform, applications can be made on their website.

For more information about Physicl and its offerings, please visit Physicl's website.

In conclusion, Physicl is not just launching a product; it’s establishing a new standard for how AI interacts with the physical world, setting the stage for the future of intelligent robotics and automated systems.