Nvidia is making significant strides in developing robotics and various industrial AI applications with the introduction of its Omniverse platform, and recently, Mega, an Omniverse Blueprint framework designed for creating digital twins to efficiently manage these applications. Additionally, Nvidia is investing in startups focused on digital twin technology to help propel this initiative.
MetAI, based in Taiwan, has introduced a model capable of swiftly producing “SimReady” (simulation-ready) digital twins via AI and 3D technology. This process transforms CAD files into functional 3D environments within mere minutes.
Nvidia has supported MetAI in its initial funding round, contributing to a $4 million seed round, marking the chip giant’s first investment in a Taiwanese startup. Other contributors to this funding round include a blend of strategic and financial investors, such as Kenmec Mechanical Engineering, Solomo Technology, SparkLabs Taiwan, Addin Ventures, and Upstream Ventures.
The forthcoming phase of AI, termed generative physical AI, depends on accurately simulated environments to train and validate robots for autonomous systems, facilitating operational AI development prior to deployment. MetAI asserts that the digital twins it creates will play a pivotal role in this process.
Daniel Yu, CEO and co-founder of MetAI, noted that digital twins have historically posed a significant barrier for physical AI due to the long development timelines, which could span months or even years.
MetAI concentrates on AI-driven digital twins specifically designed for advanced semiconductor fabrication, intelligent warehouses, and automation. It also generates synthetic data within AI-enhanced digital twin environments.
Renton Hsu, Yu’s co-founder and the CTO of MetAI, has expertise in 3D engineering and AI. He began his journey with digital twins while developing enterprise AI software applications, where they served as effective solutions in cases where clients lacked sufficient data for system training. This experience led him to the idea of applying this methodology to 3D systems, merging 3D technology with AI to formulate synthetic AI and 3D applications. Hsu collaborated with Yu, who has a background in digital transformation projects, and co-founder Dave Liu, COO, to establish MetAI.
This breakthrough achievement earned them first place in a competition hosted by Nvidia, resulting in Hsu being recognised as a “Jetson AI ambassador” for the country.
MetAI faces competition from numerous large and small corporations that have developed digital twin technologies for manufacturing, including Siemens Digital Industries, Dassault Systemes, Hexagon AB, Duality AI, and Intagles. In the realm of synthetic data, companies such as Sky Engine and Scale AI are among the many competitors.
MetAI believes its approach stands out in the industry.
Daniel Yu explained that while other companies focus on operational efficiencies or IoT integrations, MetAI employs generative models and AI-driven frameworks to produce digital twins aimed at training physical AI for real-world applications. This strategy not only speeds up the digital twin creation process but also guarantees their practical application in advanced automation systems, thereby bridging the divide between simulation and reality.
MetAI distinguishes itself by generating artificial data within its AI-enabled digital twin environments. Yu highlighted that this capability allows users to create synthetic data tailored to specific operational needs, enhancing AI training and validation. Yu emphasised that rather than producing isolated datasets, MetAI constructs dynamic virtual worlds—realistic virtual environments that mimic real-world operation.
This two-year-old startup, offering products ranging from vertical AI agents to digital twins, has secured several customers and is already generating revenue through partnerships with enterprises in the manufacturing and automation sectors. This year, it anticipates earning $3 million from a single project, according to Yu. Revenue streams include project-based income, product subscriptions, and licensing fees from ongoing developments.
Nico Caprez, corporate development manager at Nvidia, remarked that MetAI’s integration with NVIDIA Omniverse signifies a crucial advancement for industrial digital twins and physical AI in simulations. The ability to generate scalable environments for AI training could potentially establish a new benchmark across industries, from manufacturing to robotics.
In 2023, MetAI partnered with Kenmec to develop digital twins for automated warehouses. Their technology reportedly reduced the time required for warehouse digital twin simulations from thousands of hours to a mere 3 minutes, leading to substantial cost savings in operational and verification tasks.
With the recent funding, MetAI plans to enhance its R&D team to accelerate development and implement its market strategies in response to rising demand. Furthermore, the Taiwan-based startup aims to open a U.S. office and move its headquarters in the latter half of 2025, according to Yu.
Dave Liu explained that Taiwan serves as an experimental base where they collaborate with industry leaders to integrate in-depth vertical knowledge into their models, ensuring their solutions are both effective and scalable. Given Taiwan’s size and the demand for simulation-based solutions, the startup is targeting the U.S. market due to high labour costs and operational complexities. Their expansion strategy will provide both point solutions and comprehensive offerings, such as SaaS products and vertical AI agents designed for quick deployment in real-world scenarios across these sectors.