#United States #San Jose #SiMa.ai #Physical AI #Modalix

SiMa.ai Unveils Next-Gen Modalix™ Platform for Physical AI

SiMa.ai, known for its innovations in Physical AI, has recently announced a trio of pivotal product launches aimed at advancing the scalability of Physical AI applications. The centerpiece of this initiative is the release of their next-generation Machine Learning System on a Chip (MLSoC), the Modalix™, alongside a System-on-Module (SoM) and a development kit that are set to streamline production processes.

Introducing Modalix™: A Game Changer for Physical AI

The Modalix™ is engineered to be at the forefront of the Physical AI industry. It boasts exceptional performance and accuracy while maintaining low power consumption—operating under 10 watts. The Arm-based architecture of Modalix is not only flexible but also equipped with a native Generation AI stack that facilitates real-time perception and natural language processing. This makes it exceptionally suitable for diverse applications ranging from robotics and automotive to industrial automation, aerospace, and even healthcare.

Ami Badani, Chief Marketing Officer at Arm, emphasized the platform's innovative capabilities. He noted that the introduction of Modalix integrates AI and LLM functionalities directly into Physical AI at the edge, revolutionizing various industry standards for efficiency and sustainability.

Accelerating Development with Synopsys

To ensure swift development and flawless production, SiMa.ai partnered with Synopsys, employing their elite AI-powered Electronic Design Automation (EDA) suite. This collaboration has led to the production of bug-free A0 silicon, significantly enhancing SiMa.ai's confidence in launching its new products. Ravi Subramanian, Chief Product Management Officer at Synopsys, pointed out that the successful debut of the Modalix reflects their joint commitment to pushing boundaries in AI and chip design.

Collaboration with TSMC for Reliability

The Modalix leverages TSMC's advanced N6 process technology, ensuring its alignment with rigorous power, thermal, and reliability standards necessary for embedded applications. According to Sajiv Dalal, President of TSMC North America, this partnership represents an important step towards meeting the escalating demands for Physical AI solutions.

Features of the Modalix SoM and DevKit

The newly released Modalix SoM is designed for compatibility with leading GPU SoMs, providing a simple drop-in replacement for existing systems. It comes fully equipped with integrated MIPI, memory, and essential I/O features needed to expand the capabilities of Physical AI. Philipp Baechtold, CEO of Enclustra, highlighted the qualities of their joint SoM, underscoring its efficiency as a powerful plug-and-play platform.

Moreover, the Modalix ecosystem is complemented by the Palette™ software, which supports leading Machine Learning frameworks, enabling developers to transition from concepts to production seamlessly—streamlining the development timeline considerably.

LLiMa: Empowering LLM Deployment

SiMa.ai also introduced LLiMa, a unified, on-device framework that allows for the execution of LLMs and VLMs without a cloud connection. This innovative framework promotes robust functionality such as agent-2-agent systems and Retrieval-Augmented Generation (RAG), all natively executed on-device for optimal performance.

With commercial-grade units now available, the 8GB SoM starts at a competitive price of $349, while the 32GB variant is at $599. The DevKit is priced at $1,499, making the transition to scalable Physical AI both accessible and cost-effective.

Conclusion

The launch of Modalix and its associated products marks a significant milestone for SiMa.ai and the Physical AI landscape. As the demand for integrated AI solutions grows, SiMa.ai is poised to lead the charge by delivering cutting-edge hardware and software that enable faster, more intelligent systems across various industries. For those interested in learning more, additional information is available at www.sima.ai.