Jensen Huang took the stage at GTC Taipei on June 1 and did what he does best: name-dropped partners while making the case that AI is eating every industry on the planet. This time, the Nvidia CEO singled out SK Telecom as a key collaborator in bringing physical AI to semiconductor manufacturing.

The core of the announcement: SK Telecom has used Nvidia’s Omniverse platform to build a digital twin of SK Hynix’s semiconductor fabrication facility. In English, they created a detailed virtual replica of an actual chip factory, one that can simulate production processes, test optimizations, and catch problems before they cost real money on a real production line.

What SK Telecom actually built

The digital twin project feeds directly into SK Hynix’s Autonomous Factory 2030 initiative, an ambitious plan to make semiconductor fabs largely self-running within the next four years.

SK Telecom brought its own proprietary technology to the table, something it calls “Agentic Digital Twin Modeling.” The system automates the messy, time-consuming process of integrating and processing the massive amounts of data needed to make a virtual factory actually useful. Building a digital twin isn’t just about creating a 3D model that looks pretty. It requires ingesting equipment specs, sensor feeds, workflow data, and environmental conditions, then stitching them together into something that behaves like the real thing.

For Nvidia, this isn’t just a customer win. It’s a validation of its Omniverse Agent Toolkit in a real-world industrial setting.

The partnership itself isn’t brand new. SK Telecom and Nvidia announced a collaboration back in October 2025 aimed at building an AI factory powered by more than 50,000 Nvidia GPUs. The initial phase of that AI factory is expected to come online by late 2027. The GTC Taipei showcase represents the first major public demonstration of what that partnership is producing.

Why semiconductor digital twins matter right now

SK Hynix makes the memory chips that power the AI boom. Its high-bandwidth memory (HBM) products are essential components in the very Nvidia GPUs that train large language models and run inference workloads. The company recently joined the $1 trillion market cap club, driven almost entirely by AI-fueled demand for its products.

SK Telecom’s role here is worth understanding. The company is South Korea’s largest wireless carrier, but it has been aggressively diversifying into AI infrastructure and enterprise services. Partnering with Nvidia on industrial AI positions it as more than a pipe provider. It becomes a technology integrator capable of deploying complex AI systems in mission-critical environments.

What this means for investors

For SK Telecom, the partnership provides credibility in the enterprise AI space that a telecom company typically struggles to earn. The AI factory with 50,000-plus Nvidia GPUs, expected to see its first phase completed by late 2027, represents a significant capital commitment and a bet that demand for AI compute in South Korea and the broader Asia-Pacific region will justify the investment.

Investors watching this space should pay attention to whether the late 2027 AI factory milestone arrives on schedule, as that will be the first concrete test of whether the partnership can deliver on its promises at scale.