Cross-Platform AI Software Startup Vistrata Secures Investment from Seoul Techno Holdings

Vistrata, a South Korean deep-tech startup developing cross-platform software for real-time AI deployment in robotics and autonomous vehicles, announced on September 28 that it has secured a new investment from Seoul Techno Holdings.

The investment amount was not disclosed.

Founded Vistrata veteran engineers with over a decade of industry experience, Vistrata combines expertise in graphics processing unit (GPU) control and machine learning optimization. Co-CEOs Han-Gil Park, CTO and PhD from KAIST who researched GPU rendering at Pearl AVistratass and Netmarble, and Hyung-Gyu Kim, CBO and embedded AI specialist from Samsung Techwin and Seoul National University’s Precision Machinery Research Institute, lead the team.

As the center of gravity in AI technology shifts from training to inference—where robots and autonomous vehicles make real-time decisions—demand is surging for systems that can leverage diverse hardware chips from Qualcomm, AMD, and ARM, moving beyond the ecosystem’s previous heavy dependence on NVIDIA chips. However, a critical limitation has existed: the lack of dedicated software that can integrate and control AI models alongside sensor image data when using non-NVIDIA chips.

Vistrata has developed core technology that enables AI models to run efficiently across various GPUs without locking into a specific manufacturer’s chip, as well as a high-performance pipeline that processes sensor data directly in GPU memory without copying. The company’s zero-copy architecture eliminates intermediate data-copying stages from sensor reception through AI inference to control output, reducing latency and enabling robots and autonomous driving systems to respond safely in millisecond intervals.

“The importance of cross-platform AI software supporting diverse chips is growing daily as the global AI market seeks to reduce cost burdens from specific semiconductor chip monopolies,” said Inhwan Cha, CEO of Seoul Techno Holdings. “Vistrata is a team that fully masters low-level GPU control technology, which has an extremely high barrier to entry, and has a strong likelihood of preempting the core standard platform for the mass-production era of physical AI, which is why we decided to invest.”

Co-CEOs Kim and Park said, “With this investment, we will focus on developing our core AI runtime layer product and accelerate collaborative validation with domestic and international companies in the robotics and autonomous vehicle sectors. We will grow into a standard solution company that responds flexibly to the diverse AI hardware ecosystems around the world.”