NVIDIA-backed Met-AI shows you how to train your robot

NVIDIA-backed MetAI is accelerating the development of AI-driven industrial simulation technology in Taiwan.

In January 2025, NVIDIA made its first direct investment in a Taiwanese startup, backing MetAI—a company focused on accelerating the development of photorealistic, industrial-grade digital twins. The investment, though quiet, drew significant attention from the global technology and investment sectors, given MetAI’s ambition to streamline the creation of high-fidelity 3D models for industrial use.

Founded to advance robotic training through immersive virtual environments, MetAI positions itself at the intersection of AI, simulation, and automation. Its alignment with NVIDIA’s broader industrial metaverse strategy reflects a growing emphasis on applied AI in manufacturing and signals Taiwan’s expanding relevance in next-generation digital infrastructure.

From Taipei to everywhere else

Founded in 2023 and headquartered in Taipei, MetAI specializes in accelerating the creation of high-fidelity digital twins—precise virtual representations of physical environments, machinery, or products. By leveraging AI and proprietary technologies, MetAI compresses traditionally lengthy workflows into minutes, converting CAD files or sensor data into photorealistic, simulation-ready 3D models.

This capability has broad industrial applications, enabling manufacturers to design production lines swiftly, energy companies to simulate infrastructure, and logistics firms to optimize operations. CEO Daniel Yu (余泰萬) highlights, at an interview, that key sectors within Taiwan’s economy—aviation, automotive, semiconductors, and precision manufacturing—stand to gain substantially, as real-time virtualization becomes critical for operational efficiency and competitive advantage.

Why NVIDIA took notice

NVIDIA, a longstanding leader in AI computing and industrial simulation, has systematically built partnerships to advance its Omniverse platform—a real-time, physics-accurate simulation environment aimed at transforming industrial workflows. MetAI’s MetGen platform, which generates simulation-ready 3D assets from real-world visual data, aligns directly with this strategy by streamlining asset creation and embedding vertical domain expertise into NVIDIA’s broader cross-industry ecosystem.

This alignment was formalized through a US$4 million seed round co-led by SparkLabs Taiwan and joined by NVIDIA’s venture arm, along with strategic investors including Kenmec Mechanical Engineering and Solomon Technology. Beyond capital, NVIDIA has enrolled MetAI in its Inception Program, providing access to advanced computing infrastructure and technical collaboration—resources critical to scaling the company’s AI-native 3D generation capabilities within industrial contexts.


From obscurity to industry spotlight

MetAI began with a focused effort to refine its Real-to-Sim technology through collaborations with local Taiwanese manufacturers and robotics firms. These early-stage projects allowed the company to validate the accuracy and efficiency of its AI-generated simulations in real industrial settings.

Its turning point came when internal benchmarks showed that its models could outperform traditional CAD-based simulations in both speed and visual fidelity. As demand grew globally for scalable virtual environments to train AI systems, MetAI’s offering gained strategic relevance. The company now serves clients in smart manufacturing, energy infrastructure, and autonomous systems, with partners including TSMC and Wistron—signaling growing commercial momentum despite the absence of disclosed revenue data.

A closer look

Digital twin creation is often mistaken for a basic rendering task, but it involves a complex multi-modal process—requiring accurate data capture, synthesis, geometric reconstruction, texture generation, and physics-based integration. MetAI addresses this through a hybrid architecture combining neural radiance fields (NeRF), generative adversarial networks (GANs), and proprietary domain-adaptation algorithms, enabling it to convert noisy, real-world inputs such as factory floor video into clean, simulation-ready assets with minimal manual processing.

A key differentiator is MetAI’s Real-to-Sim and Sim-to-Real pipeline, which enables AI models to train in highly realistic virtual environments and deploy more effectively in physical settings. This significantly reduces the cost and friction of bridging simulation with real-world application—an advantage for sectors like autonomous systems and precision manufacturing.

As Daniel Yu put it, the same technology could help manufacturers relocate production across borders with speed and accuracy amid trade or geopolitical disruptions. With the global digital twin market projected to exceed US$180 billion by 2030, MetAI’s ability to automate asset generation positions it to capture a significant share of this rapidly expanding sector.

 Taiwan’s evolving innovation ecosystem 

Taiwan’s global reputation in semiconductors has long overshadowed its emerging AI startup ecosystem. However, MetAI’s rapid ascent, underscored by direct investment from NVIDIA, signals a potential turning point, positioning Taiwan as a viable hub for deep tech entrepreneurship beyond hardware manufacturing.

This shift is supported by structural advantages. Taiwan’s dense industrial base provides fertile ground for real-world testing and iterative development, while its engineering workforce offers both technical sophistication and cost efficiency.

According to MetAI CEO Daniel Yu, these conditions form a critical feedback loop between AI developers and industrial adopters. Institutions like the Taiwan AI Academy—where MetAI CTO Renton Hsu (徐嘉呈) previously worked—have also played a foundational role in building the talent infrastructure necessary to support such globally oriented innovation.


Teething pain

Despite its strong technological foundation, MetAI faces a range of challenges common to deep tech ventures. Enterprise adoption cycles in industrial sectors are often prolonged, and integrating advanced AI into legacy systems remains complex and resource-intensive. The digital twin market itself is becoming increasingly competitive, with established players such as Siemens, Dassault Systèmes, and Unity Technologies expanding their capabilities.

The company must also remain vigilant against the rapid pace of innovation, particularly as open-source tools for asset generation become more accessible. Additionally, as a Taiwanese firm operating within US-aligned ecosystems, MetAI must navigate the geopolitical sensitivities of cross-border collaboration in a tense US-China tech landscape.

However, NVIDIA’s backing provides both validation and strategic alignment. Deeper integration into the Omniverse platform could offer MetAI a durable competitive advantage, anchoring it within a broader ecosystem that may prove difficult for competitors to displace.

 Industrial AI’s inflection point 

MetAI’s trajectory reflects a broader inflection point. The industrial metaverse is transitioning from a theoretical vision to practical deployment. For years, the integration of physical and virtual systems remained aspirational, constrained by high development costs and complex workflows—even for major players like Meta. The rise of generative AI, capable of converting raw sensor inputs into simulation-ready digital environments, is now removing many of those barriers.

NVIDIA’s investment signals that the next wave of AI adoption will prioritize industrial applications—focusing on productivity, infrastructure, and systems-level optimization—rather than consumer-facing tools alone. While generative AI for text and images garners mainstream attention, the deeper transformation may occur in sectors where AI-driven simulation reshapes how global industries design, test, and operate.

Moving on

In the competitive landscape of AI innovation, securing investment from a global leader like NVIDIA is a rare achievement. For MetAI, this partnership represents not only financial support but also a strong endorsement of its technological vision and operational capabilities, opening pathways to international markets.

Sustaining this momentum will require MetAI to maintain its technical leadership, navigate an evolving competitive environment, and effectively leverage NVIDIA’s ecosystem. If successful, this collaboration could mark a pivotal chapter in Taiwan’s AI development, underscoring the convergence of innovation, strategic positioning, and timely opportunity.